experiments/STT
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge remote-tracking branch 'upstream/master' into update-r0.8

Browse Source
This commit is contained in:
Alexandre Lissy 2020-07-27 10:29:03 +02:00
commit 39ef96b533
79 changed files with 3111 additions and 431 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

92
bin/build_sdb.py
View File

@ -1,92 +0,0 @@
#!/usr/bin/env python
"""
Tool for building Sample Databases (SDB files) from DeepSpeech CSV files and other SDB files
Use "python3 build_sdb.py -h" for help
"""
import argparse
import progressbar
from deepspeech_training.util.audio import (
AUDIO_TYPE_PCM,
AUDIO_TYPE_OPUS,
AUDIO_TYPE_WAV,
change_audio_types,
)
from deepspeech_training.util.downloader import SIMPLE_BAR
from deepspeech_training.util.sample_collections import (
DirectSDBWriter,
samples_from_sources,
)
from deepspeech_training.util.augmentations import (
parse_augmentations,
apply_sample_augmentations,
SampleAugmentation
)
AUDIO_TYPE_LOOKUP = {"wav": AUDIO_TYPE_WAV, "opus": AUDIO_TYPE_OPUS}
def build_sdb():
audio_type = AUDIO_TYPE_LOOKUP[CLI_ARGS.audio_type]
augmentations = parse_augmentations(CLI_ARGS.augment)
if any(not isinstance(a, SampleAugmentation) for a in augmentations):
print("Warning: Some of the augmentations cannot be applied by this command.")
with DirectSDBWriter(
CLI_ARGS.target, audio_type=audio_type, labeled=not CLI_ARGS.unlabeled
) as sdb_writer:
samples = samples_from_sources(CLI_ARGS.sources, labeled=not CLI_ARGS.unlabeled)
num_samples = len(samples)
if augmentations:
samples = apply_sample_augmentations(samples, audio_type=AUDIO_TYPE_PCM, augmentations=augmentations)
bar = progressbar.ProgressBar(max_value=num_samples, widgets=SIMPLE_BAR)
for sample in bar(
change_audio_types(samples, audio_type=audio_type, bitrate=CLI_ARGS.bitrate, processes=CLI_ARGS.workers)
):
sdb_writer.add(sample)
def handle_args():
parser = argparse.ArgumentParser(
description="Tool for building Sample Databases (SDB files) "
"from DeepSpeech CSV files and other SDB files"
)
parser.add_argument(
"sources",
nargs="+",
help="Source CSV and/or SDB files - "
"Note: For getting a correctly ordered target SDB, source SDBs have to have their samples "
"already ordered from shortest to longest.",
)
parser.add_argument("target", help="SDB file to create")
parser.add_argument(
"--audio-type",
default="opus",
choices=AUDIO_TYPE_LOOKUP.keys(),
help="Audio representation inside target SDB",
)
parser.add_argument(
"--bitrate",
type=int,
help="Bitrate for lossy compressed SDB samples like in case of --audio-type opus",
)
parser.add_argument(
"--workers", type=int, default=None, help="Number of encoding SDB workers"
)
parser.add_argument(
"--unlabeled",
action="store_true",
help="If to build an SDB with unlabeled (audio only) samples - "
"typically used for building noise augmentation corpora",
)
parser.add_argument(
"--augment",
action='append',
help="Add an augmentation operation",
)
return parser.parse_args()
if __name__ == "__main__":
CLI_ARGS = handle_args()
build_sdb()

111
bin/data_set_tool.py Executable file
View File

@ -0,0 +1,111 @@
#!/usr/bin/env python
'''
Tool for building a combined SDB or CSV sample-set from other sets
Use 'python3 data_set_tool.py -h' for help
'''
import sys
import argparse
import progressbar
from pathlib import Path
from deepspeech_training.util.audio import (
AUDIO_TYPE_PCM,
AUDIO_TYPE_OPUS,
AUDIO_TYPE_WAV,
change_audio_types,
)
from deepspeech_training.util.downloader import SIMPLE_BAR
from deepspeech_training.util.sample_collections import (
CSVWriter,
DirectSDBWriter,
samples_from_sources,
)
from deepspeech_training.util.augmentations import (
parse_augmentations,
apply_sample_augmentations,
SampleAugmentation
)
AUDIO_TYPE_LOOKUP = {'wav': AUDIO_TYPE_WAV, 'opus': AUDIO_TYPE_OPUS}
def build_data_set():
audio_type = AUDIO_TYPE_LOOKUP[CLI_ARGS.audio_type]
augmentations = parse_augmentations(CLI_ARGS.augment)
if any(not isinstance(a, SampleAugmentation) for a in augmentations):
print('Warning: Some of the specified augmentations will not get applied, as this tool only supports '
'overlay, codec, reverb, resample and volume.')
extension = Path(CLI_ARGS.target).suffix.lower()
labeled = not CLI_ARGS.unlabeled
if extension == '.csv':
writer = CSVWriter(CLI_ARGS.target, absolute_paths=CLI_ARGS.absolute_paths, labeled=labeled)
elif extension == '.sdb':
writer = DirectSDBWriter(CLI_ARGS.target, audio_type=audio_type, labeled=labeled)
else:
print('Unknown extension of target file - has to be either .csv or .sdb')
sys.exit(1)
with writer:
samples = samples_from_sources(CLI_ARGS.sources, labeled=not CLI_ARGS.unlabeled)
num_samples = len(samples)
if augmentations:
samples = apply_sample_augmentations(samples, audio_type=AUDIO_TYPE_PCM, augmentations=augmentations)
bar = progressbar.ProgressBar(max_value=num_samples, widgets=SIMPLE_BAR)
for sample in bar(change_audio_types(
samples,
audio_type=audio_type,
bitrate=CLI_ARGS.bitrate,
processes=CLI_ARGS.workers)):
writer.add(sample)
def handle_args():
parser = argparse.ArgumentParser(
description='Tool for building a combined SDB or CSV sample-set from other sets'
)
parser.add_argument(
'sources',
nargs='+',
help='Source CSV and/or SDB files - '
'Note: For getting a correctly ordered target set, source SDBs have to have their samples '
'already ordered from shortest to longest.',
)
parser.add_argument(
'target',
help='SDB or CSV file to create'
)
parser.add_argument(
'--audio-type',
default='opus',
choices=AUDIO_TYPE_LOOKUP.keys(),
help='Audio representation inside target SDB',
)
parser.add_argument(
'--bitrate',
type=int,
help='Bitrate for lossy compressed SDB samples like in case of --audio-type opus',
)
parser.add_argument(
'--workers', type=int, default=None, help='Number of encoding SDB workers'
)
parser.add_argument(
'--unlabeled',
action='store_true',
help='If to build an SDB with unlabeled (audio only) samples - '
'typically used for building noise augmentation corpora',
)
parser.add_argument(
'--absolute-paths',
action='store_true',
help='If to reference samples by their absolute paths when writing CSV files',
)
parser.add_argument(
'--augment',
action='append',
help='Add an augmentation operation',
)
return parser.parse_args()
if __name__ == '__main__':
CLI_ARGS = handle_args()
build_data_set()

2
bin/play.py
View File

@ -1,7 +1,7 @@
#!/usr/bin/env python #!/usr/bin/env python
""" """
Tool for playing (and augmenting) single samples or samples from Sample Databases (SDB files) and DeepSpeech CSV files Tool for playing (and augmenting) single samples or samples from Sample Databases (SDB files) and DeepSpeech CSV files
Use "python3 build_sdb.py -h" for help Use "python3 play.py -h" for help
""" """
import os import os

2
bin/run-tc-ldc93s1_checkpoint_sdb.sh
View File

@ -13,7 +13,7 @@ fi;
if [ ! -f "${ldc93s1_dir}/ldc93s1.sdb" ]; then if [ ! -f "${ldc93s1_dir}/ldc93s1.sdb" ]; then
echo "Converting LDC93S1 example data, saving to ${ldc93s1_sdb}." echo "Converting LDC93S1 example data, saving to ${ldc93s1_sdb}."
python -u bin/build_sdb.py ${ldc93s1_csv} ${ldc93s1_sdb} python -u bin/data_set_tool.py ${ldc93s1_csv} ${ldc93s1_sdb}
fi; fi;
# Force only one visible device because we have a single-sample dataset # Force only one visible device because we have a single-sample dataset

2
bin/run-tc-ldc93s1_new_sdb.sh
View File

@ -16,7 +16,7 @@ fi;
if [ ! -f "${ldc93s1_dir}/ldc93s1.sdb" ]; then if [ ! -f "${ldc93s1_dir}/ldc93s1.sdb" ]; then
echo "Converting LDC93S1 example data, saving to ${ldc93s1_sdb}." echo "Converting LDC93S1 example data, saving to ${ldc93s1_sdb}."
python -u bin/build_sdb.py ${ldc93s1_csv} ${ldc93s1_sdb} python -u bin/data_set_tool.py ${ldc93s1_csv} ${ldc93s1_sdb}
fi; fi;
# Force only one visible device because we have a single-sample dataset # Force only one visible device because we have a single-sample dataset

2
bin/run-tc-ldc93s1_new_sdb_csv.sh
View File

@ -16,7 +16,7 @@ fi;
if [ ! -f "${ldc93s1_dir}/ldc93s1.sdb" ]; then if [ ! -f "${ldc93s1_dir}/ldc93s1.sdb" ]; then
echo "Converting LDC93S1 example data, saving to ${ldc93s1_sdb}." echo "Converting LDC93S1 example data, saving to ${ldc93s1_sdb}."
python -u bin/build_sdb.py ${ldc93s1_csv} ${ldc93s1_sdb} python -u bin/data_set_tool.py ${ldc93s1_csv} ${ldc93s1_sdb}
fi; fi;
# Force only one visible device because we have a single-sample dataset # Force only one visible device because we have a single-sample dataset

14
doc/BUILDING.rst
View File

@ -77,6 +77,20 @@ You can now use Bazel to build the main DeepSpeech library, ``libdeepspeech.so``
The generated binaries will be saved to ``bazel-bin/native_client/``. The generated binaries will be saved to ``bazel-bin/native_client/``.
.. _build-generate-scorer-package:
Compile ``generate_scorer_package``
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Following the same setup as for ``libdeepspeech.so`` above, you can rebuild the ``generate_scorer_package`` binary by adding its target to the command line: ``//native_client:generate_scorer_package``.
Using the example from above you can build the library and that binary at the same time:
.. code-block::
bazel build --workspace_status_command="bash native_client/bazel_workspace_status_cmd.sh" --config=monolithic -c opt --copt=-O3 --copt="-D_GLIBCXX_USE_CXX11_ABI=0" --copt=-fvisibility=hidden //native_client:libdeepspeech.so //native_client:generate_scorer_package
The generated binaries will be saved to ``bazel-bin/native_client/``.
Compile Language Bindings Compile Language Bindings
^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^

3
doc/Scorer.rst
View File

@ -49,6 +49,9 @@ Afterwards you can use ``generate_scorer_package`` to generate the scorer packag
./generate_scorer_package --alphabet ../alphabet.txt --lm lm.binary --vocab vocab-500000.txt \ ./generate_scorer_package --alphabet ../alphabet.txt --lm lm.binary --vocab vocab-500000.txt \
--package kenlm.scorer --default_alpha 0.931289039105002 --default_beta 1.1834137581510284 --package kenlm.scorer --default_alpha 0.931289039105002 --default_beta 1.1834137581510284
The ``generate_scorer_package`` binary is part of the released ``native_client.tar.xz``. If for some reason you need to rebuild it,
please refer to how to :ref:`build-generate-scorer-package`.
Building your own scorer Building your own scorer
------------------------ ------------------------

15
doc/TRAINING.rst
View File

@ -124,7 +124,7 @@ It will also add the following ``.csv`` files:
* ``clips/dev.csv`` * ``clips/dev.csv``
* ``clips/test.csv`` * ``clips/test.csv``
All entries in these CSV files refer to their samples by absolute paths. So moving this sub-directory would require another import or tweaking the CSV files accordingly. Entries in CSV files can refer to samples by their absolute or relative paths. Here, the importer produces relative paths.
To use Common Voice data during training, validation and testing, you pass (comma separated combinations of) their filenames into ``--train_files``\ , ``--dev_files``\ , ``--test_files`` parameters of ``DeepSpeech.py``. To use Common Voice data during training, validation and testing, you pass (comma separated combinations of) their filenames into ``--train_files``\ , ``--dev_files``\ , ``--test_files`` parameters of ``DeepSpeech.py``.
@ -287,6 +287,8 @@ UTF-8 mode
DeepSpeech includes a UTF-8 operating mode which can be useful to model languages with very large alphabets, such as Chinese Mandarin. For details on how it works and how to use it, see :ref:`decoder-docs`. DeepSpeech includes a UTF-8 operating mode which can be useful to model languages with very large alphabets, such as Chinese Mandarin. For details on how it works and how to use it, see :ref:`decoder-docs`.
.. _training-data-augmentation:
Augmentation Augmentation
^^^^^^^^^^^^ ^^^^^^^^^^^^
@ -496,7 +498,7 @@ Example training with all augmentations:
[...] [...]
The ``bin/play.py`` tool also supports ``--augment`` parameters (for sample domain augmentations) and can be used for experimenting with different configurations. The ``bin/play.py`` and ``bin/data_set_tool.py`` tools also support ``--augment`` parameters (for sample domain augmentations) and can be used for experimenting with different configurations or creating augmented data sets.
Example of playing all samples with reverberation and maximized volume: Example of playing all samples with reverberation and maximized volume:
@ -510,3 +512,12 @@ Example simulation of the codec augmentation of a wav-file first at the beginnin
bin/play.py --augment codec[p=0.1,bitrate=48000:16000] --clock 0.0 test.wav bin/play.py --augment codec[p=0.1,bitrate=48000:16000] --clock 0.0 test.wav
bin/play.py --augment codec[p=0.1,bitrate=48000:16000] --clock 1.0 test.wav bin/play.py --augment codec[p=0.1,bitrate=48000:16000] --clock 1.0 test.wav
Example of creating a pre-augmented test set:
.. code-block:: bash
bin/data_set_tool.py \
--augment overlay[source=noise.sdb,layers=1,snr=20~10] \
--augment resample[rate=12000:8000~4000] \
test.sdb test-augmented.sdb

1
native_client/BUILD
View File

@ -130,6 +130,7 @@ tf_cc_shared_object(
}) + tflite_copts(), }) + tflite_copts(),
linkopts = select({ linkopts = select({
"//tensorflow:macos": [], "//tensorflow:macos": [],
"//tensorflow:ios": ["-fembed-bitcode"],
"//tensorflow:linux_x86_64": LINUX_LINKOPTS, "//tensorflow:linux_x86_64": LINUX_LINKOPTS,
"//tensorflow:rpi3": LINUX_LINKOPTS, "//tensorflow:rpi3": LINUX_LINKOPTS,
"//tensorflow:rpi3-armv8": LINUX_LINKOPTS, "//tensorflow:rpi3-armv8": LINUX_LINKOPTS,

10
native_client/client.cc
View File

@ -12,7 +12,11 @@
#include <sstream> #include <sstream>
#include <string> #include <string>
#if defined(__ANDROID__) || defined(_MSC_VER) #ifdef __APPLE__
#include <TargetConditionals.h>
#endif
#if defined(__ANDROID__) || defined(_MSC_VER) || TARGET_OS_IPHONE
#define NO_SOX #define NO_SOX
#endif #endif
@ -244,7 +248,7 @@ GetAudioBuffer(const char* path, int desired_sample_rate)
sox_false // Reverse endianness sox_false // Reverse endianness
}; };
#ifdef __APPLE__ #if TARGET_OS_OSX
// It would be preferable to use sox_open_memstream_write here, but OS-X // It would be preferable to use sox_open_memstream_write here, but OS-X
// doesn't support POSIX 2008, which it requires. See Issue #461. // doesn't support POSIX 2008, which it requires. See Issue #461.
// Instead, we write to a temporary file. // Instead, we write to a temporary file.
@ -348,7 +352,7 @@ GetAudioBuffer(const char* path, int desired_sample_rate)
fclose(wave); fclose(wave);
#endif // NO_SOX #endif // NO_SOX
#ifdef __APPLE__ #if TARGET_OS_OSX
res.buffer_size = (size_t)(output->olength * 2); res.buffer_size = (size_t)(output->olength * 2);
res.buffer = (char*)malloc(sizeof(char) * res.buffer_size); res.buffer = (char*)malloc(sizeof(char) * res.buffer_size);
FILE* output_file = fopen(output_name, "rb"); FILE* output_file = fopen(output_name, "rb");

19
native_client/ctcdecode/__init__.py
View File

@ -48,15 +48,33 @@ class Alphabet(swigwrapper.Alphabet):
raise ValueError('Alphabet initialization failed with error code 0x{:X}'.format(err)) raise ValueError('Alphabet initialization failed with error code 0x{:X}'.format(err))
def CanEncodeSingle(self, input): def CanEncodeSingle(self, input):
'''
Returns true if the single character/output class has a corresponding label
in the alphabet.
'''
return super(Alphabet, self).CanEncodeSingle(input.encode('utf-8')) return super(Alphabet, self).CanEncodeSingle(input.encode('utf-8'))
def CanEncode(self, input): def CanEncode(self, input):
'''
Returns true if the entire string can be encoded into labels in this
alphabet.
'''
return super(Alphabet, self).CanEncode(input.encode('utf-8')) return super(Alphabet, self).CanEncode(input.encode('utf-8'))
def EncodeSingle(self, input): def EncodeSingle(self, input):
'''
Encode a single character/output class into a label. Character must be in
the alphabet, this method will assert that. Use `CanEncodeSingle` to test.
'''
return super(Alphabet, self).EncodeSingle(input.encode('utf-8')) return super(Alphabet, self).EncodeSingle(input.encode('utf-8'))
def Encode(self, input): def Encode(self, input):
'''
Encode a sequence of character/output classes into a sequence of labels.
Characters are assumed to always take a single Unicode codepoint.
Characters must be in the alphabet, this method will assert that. Use
`CanEncode` and `CanEncodeSingle` to test.
'''
# Convert SWIG's UnsignedIntVec to a Python list # Convert SWIG's UnsignedIntVec to a Python list
res = super(Alphabet, self).Encode(input.encode('utf-8')) res = super(Alphabet, self).Encode(input.encode('utf-8'))
return [el for el in res] return [el for el in res]
@ -66,6 +84,7 @@ class Alphabet(swigwrapper.Alphabet):
return res.decode('utf-8') return res.decode('utf-8')
def Decode(self, input): def Decode(self, input):
'''Decode a sequence of labels into a string.'''
res = super(Alphabet, self).Decode(input) res = super(Alphabet, self).Decode(input)
return res.decode('utf-8') return res.decode('utf-8')

22
native_client/definitions.mk
View File

@ -101,6 +101,20 @@ NODE_PLATFORM_TARGET := --target_arch=arm64 --target_platform=linux
TOOLCHAIN_LDD_OPTS := --root $(RASPBIAN)/ TOOLCHAIN_LDD_OPTS := --root $(RASPBIAN)/
endif # ($(TARGET),rpi3-armv8) endif # ($(TARGET),rpi3-armv8)
ifeq ($(TARGET),ios-simulator)
CFLAGS := -isysroot $(shell xcrun -sdk iphonesimulator13.5 -show-sdk-path)
SOX_CFLAGS :=
SOX_LDFLAGS :=
LDFLAGS :=
endif
ifeq ($(TARGET),ios-arm64)
CFLAGS := -target arm64-apple-ios -isysroot $(shell xcrun -sdk iphoneos13.5 -show-sdk-path)
SOX_CFLAGS :=
SOX_LDFLAGS :=
LDFLAGS :=
endif
# -Wl,--no-as-needed is required to force linker not to evict libs it thinks we # -Wl,--no-as-needed is required to force linker not to evict libs it thinks we
# dont need ; will fail the build on OSX because that option does not exists # dont need ; will fail the build on OSX because that option does not exists
ifeq ($(OS),Linux) ifeq ($(OS),Linux)
@ -108,9 +122,13 @@ LDFLAGS_NEEDED := -Wl,--no-as-needed
LDFLAGS_RPATH := -Wl,-rpath,\$$ORIGIN LDFLAGS_RPATH := -Wl,-rpath,\$$ORIGIN
endif endif
ifeq ($(OS),Darwin) ifeq ($(OS),Darwin)
CXXFLAGS += -stdlib=libc++ -mmacosx-version-min=10.10 CXXFLAGS += -stdlib=libc++
LDFLAGS_NEEDED := -stdlib=libc++ -mmacosx-version-min=10.10 LDFLAGS_NEEDED := -stdlib=libc++
LDFLAGS_RPATH := -Wl,-rpath,@executable_path LDFLAGS_RPATH := -Wl,-rpath,@executable_path
ifeq ($(TARGET),host)
CXXFLAGS += -mmacosx-version-min=10.10
LDFLAGS_NEEDED += -mmacosx-version-min=10.10
endif
endif endif
CFLAGS += $(EXTRA_CFLAGS) CFLAGS += $(EXTRA_CFLAGS)

6
native_client/dotnet/DeepSpeech.sln
View File

@ -1,8 +1,8 @@
Microsoft Visual Studio Solution File, Format Version 12.00 Microsoft Visual Studio Solution File, Format Version 12.00
# Visual Studio 15 # Visual Studio Version 16
VisualStudioVersion = 15.0.28307.136 VisualStudioVersion = 16.0.30204.135
MinimumVisualStudioVersion = 10.0.40219.1 MinimumVisualStudioVersion = 10.0.40219.1
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "DeepSpeechClient", "DeepSpeechClient\DeepSpeechClient.csproj", "{56DE4091-BBBE-47E4-852D-7268B33B971F}" Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "DeepSpeechClient", "DeepSpeechClient\DeepSpeechClient.csproj", "{56DE4091-BBBE-47E4-852D-7268B33B971F}"
EndProject EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "DeepSpeechConsole", "DeepSpeechConsole\DeepSpeechConsole.csproj", "{312965E5-C4F6-4D95-BA64-79906B8BC7AC}" Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "DeepSpeechConsole", "DeepSpeechConsole\DeepSpeechConsole.csproj", "{312965E5-C4F6-4D95-BA64-79906B8BC7AC}"
EndProject EndProject

85
native_client/dotnet/DeepSpeechClient/DeepSpeechClient.csproj
View File

@ -1,63 +1,42 @@
<?xml version="1.0" encoding="utf-8"?> <Project Sdk="Microsoft.NET.Sdk">
<Project ToolsVersion="15.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
<Import Project="$(MSBuildExtensionsPath)\$(MSBuildToolsVersion)\Microsoft.Common.props" Condition="Exists('$(MSBuildExtensionsPath)\$(MSBuildToolsVersion)\Microsoft.Common.props')" />
<PropertyGroup> <PropertyGroup>
<Configuration Condition=" '$(Configuration)' == '' ">Debug</Configuration> <OutputType>Library</OutputType>
<Platform Condition=" '$(Platform)' == '' ">AnyCPU</Platform> <TargetFrameworks>net452;net46;net47;uap10.0</TargetFrameworks>
<ProjectGuid>{56DE4091-BBBE-47E4-852D-7268B33B971F}</ProjectGuid> <Platforms>x64</Platforms>
<OutputType>Library</OutputType>
<AppDesignerFolder>Properties</AppDesignerFolder>
<RootNamespace>DeepSpeechClient</RootNamespace>
<AssemblyName>DeepSpeechClient</AssemblyName>
<TargetFrameworkVersion>v4.6.2</TargetFrameworkVersion>
<FileAlignment>512</FileAlignment>
<Deterministic>true</Deterministic>
</PropertyGroup> </PropertyGroup>
<PropertyGroup Condition="'$(Configuration)|$(Platform)' == 'Debug|x64'"> <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x86'">
<DebugSymbols>true</DebugSymbols>
<OutputPath>bin\x64\Debug\</OutputPath>
<DefineConstants>DEBUG;TRACE</DefineConstants>
<AllowUnsafeBlocks>true</AllowUnsafeBlocks>
<DebugType>full</DebugType>
<PlatformTarget>x64</PlatformTarget> <PlatformTarget>x64</PlatformTarget>
<ErrorReport>prompt</ErrorReport>
<CodeAnalysisRuleSet>MinimumRecommendedRules.ruleset</CodeAnalysisRuleSet>
</PropertyGroup> </PropertyGroup>
<PropertyGroup Condition="'$(Configuration)|$(Platform)' == 'Release|x64'"> <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
<OutputPath>bin\x64\Release\</OutputPath>
<DefineConstants>TRACE</DefineConstants>
<Optimize>true</Optimize>
<DebugType>pdbonly</DebugType>
<PlatformTarget>x64</PlatformTarget>
<ErrorReport>prompt</ErrorReport>
<CodeAnalysisRuleSet>MinimumRecommendedRules.ruleset</CodeAnalysisRuleSet>
<AllowUnsafeBlocks>true</AllowUnsafeBlocks> <AllowUnsafeBlocks>true</AllowUnsafeBlocks>
</PropertyGroup> </PropertyGroup>
<ItemGroup> <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
<Reference Include="System" /> <AllowUnsafeBlocks>true</AllowUnsafeBlocks>
<Reference Include="System.Core" /> <PlatformTarget>x64</PlatformTarget>
<Reference Include="System.Xml.Linq" /> </PropertyGroup>
<Reference Include="System.Data.DataSetExtensions" />
<Reference Include="Microsoft.CSharp" /> <PropertyGroup Condition="'$(TargetFramework)' == 'uap10.0'">
<Reference Include="System.Data" /> <CopyLocalLockFileAssemblies>false</CopyLocalLockFileAssemblies>
<Reference Include="System.Net.Http" /> <NugetTargetMoniker>UAP,Version=v10.0</NugetTargetMoniker>
<Reference Include="System.Xml" /> <TargetPlatformIdentifier>UAP</TargetPlatformIdentifier>
<TargetPlatformVersion>10.0.19041.0</TargetPlatformVersion>
<TargetPlatformMinVersion>10.0.10240.0</TargetPlatformMinVersion>
<TargetFrameworkIdentifier>.NETCore</TargetFrameworkIdentifier>
<TargetFrameworkVersion>v5.0</TargetFrameworkVersion>
<DefineConstants>$(DefineConstants);WINDOWS_UWP</DefineConstants>
<LanguageTargets>$(MSBuildExtensionsPath)\Microsoft\WindowsXaml\v$(VisualStudioVersion)\Microsoft.Windows.UI.Xaml.CSharp.targets</LanguageTargets>
<ResolveAssemblyWarnOrErrorOnTargetArchitectureMismatch>None</ResolveAssemblyWarnOrErrorOnTargetArchitectureMismatch>
</PropertyGroup>
<ItemGroup Condition=" '$(TargetFramework)' == 'uap10.0' ">
<PackageReference Include="Microsoft.NETCore.UniversalWindowsPlatform " Version="5.2.2" />
</ItemGroup> </ItemGroup>
<ItemGroup> <ItemGroup>
<Compile Include="DeepSpeech.cs" /> <Folder Include="Properties\" />
<Compile Include="Enums\ErrorCodes.cs" />
<Compile Include="Interfaces\IDeepSpeech.cs" />
<Compile Include="Extensions\NativeExtensions.cs" />
<Compile Include="Models\DeepSpeechStream.cs" />
<Compile Include="Models\Metadata.cs" />
<Compile Include="Models\CandidateTranscript.cs" />
<Compile Include="Models\TokenMetadata.cs" />
<Compile Include="NativeImp.cs" />
<Compile Include="Properties\AssemblyInfo.cs" />
<Compile Include="Structs\Metadata.cs" />
<Compile Include="Structs\CandidateTranscript.cs" />
<Compile Include="Structs\TokenMetadata.cs" />
</ItemGroup> </ItemGroup>
<ItemGroup />
<Import Project="$(MSBuildToolsPath)\Microsoft.CSharp.targets" />
<PropertyGroup Condition=" '$(TargetFramework)' == 'uap10.0' ">
<DefineConstants>$(DefineConstants);NO_HTTPS</DefineConstants>
</PropertyGroup>
</Project> </Project>

36
native_client/dotnet/DeepSpeechClient/Properties/AssemblyInfo.cs
View File

@ -1,36 +0,0 @@
using System.Reflection;
using System.Runtime.CompilerServices;
using System.Runtime.InteropServices;
// General Information about an assembly is controlled through the following
// set of attributes. Change these attribute values to modify the information
// associated with an assembly.
[assembly: AssemblyTitle("DeepSpeechClient")]
[assembly: AssemblyDescription("")]
[assembly: AssemblyConfiguration("")]
[assembly: AssemblyCompany("")]
[assembly: AssemblyProduct("DeepSpeechClient")]
[assembly: AssemblyCopyright("Copyright © 2018")]
[assembly: AssemblyTrademark("")]
[assembly: AssemblyCulture("")]
// Setting ComVisible to false makes the types in this assembly not visible
// to COM components. If you need to access a type in this assembly from
// COM, set the ComVisible attribute to true on that type.
[assembly: ComVisible(false)]
// The following GUID is for the ID of the typelib if this project is exposed to COM
[assembly: Guid("56de4091-bbbe-47e4-852d-7268b33b971f")]
// Version information for an assembly consists of the following four values:
//
// Major Version
// Minor Version
// Build Number
// Revision
//
// You can specify all the values or you can default the Build and Revision Numbers
// by using the '*' as shown below:
// [assembly: AssemblyVersion("1.0.*")]
[assembly: AssemblyVersion("1.0.0.0")]
[assembly: AssemblyFileVersion("1.0.0.0")]

4
native_client/swift/.gitignore vendored Normal file
View File

@ -0,0 +1,4 @@
.DS_Store
build/
xcuserdata/
/deepspeech_ios/libdeepspeech.dylib

499
native_client/swift/deepspeech_ios.xcodeproj/project.pbxproj Normal file
View File

@ -0,0 +1,499 @@
// !$*UTF8*$!
{
archiveVersion = 1;
classes = {
};
objectVersion = 50;
objects = {
/* Begin PBXBuildFile section */
505B136B24960D550007DADA /* deepspeech_ios.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 505B136124960D550007DADA /* deepspeech_ios.framework */; };
505B137224960D550007DADA /* deepspeech_ios.h in Headers */ = {isa = PBXBuildFile; fileRef = 505B136424960D550007DADA /* deepspeech_ios.h */; settings = {ATTRIBUTES = (Public, ); }; };
505B137D24961AF20007DADA /* deepspeech.h in Headers */ = {isa = PBXBuildFile; fileRef = 505B137C24961AF20007DADA /* deepspeech.h */; settings = {ATTRIBUTES = (Private, ); }; };
505B137F24961BA70007DADA /* DeepSpeech.swift in Sources */ = {isa = PBXBuildFile; fileRef = 505B137E24961BA70007DADA /* DeepSpeech.swift */; };
507CD39B24B61FA100409BBB /* libdeepspeech.so in Frameworks */ = {isa = PBXBuildFile; fileRef = 507CD39A24B61FA100409BBB /* libdeepspeech.so */; };
/* End PBXBuildFile section */
/* Begin PBXContainerItemProxy section */
505B136C24960D550007DADA /* PBXContainerItemProxy */ = {
isa = PBXContainerItemProxy;
containerPortal = 505B135824960D550007DADA /* Project object */;
proxyType = 1;
remoteGlobalIDString = 505B136024960D550007DADA;
remoteInfo = deepspeech_ios;
};
/* End PBXContainerItemProxy section */
/* Begin PBXCopyFilesBuildPhase section */
505B138A249628290007DADA /* Embed Libraries */ = {
isa = PBXCopyFilesBuildPhase;
buildActionMask = 12;
dstPath = "";
dstSubfolderSpec = 10;
files = (
);
name = "Embed Libraries";
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXCopyFilesBuildPhase section */
/* Begin PBXFileReference section */
505B136124960D550007DADA /* deepspeech_ios.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = deepspeech_ios.framework; sourceTree = BUILT_PRODUCTS_DIR; };
505B136424960D550007DADA /* deepspeech_ios.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = deepspeech_ios.h; sourceTree = "<group>"; };
505B136524960D550007DADA /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
505B136A24960D550007DADA /* deepspeech_iosTests.xctest */ = {isa = PBXFileReference; explicitFileType = wrapper.cfbundle; includeInIndex = 0; path = deepspeech_iosTests.xctest; sourceTree = BUILT_PRODUCTS_DIR; };
505B137B249619C90007DADA /* deepspeech_ios.modulemap */ = {isa = PBXFileReference; lastKnownFileType = "sourcecode.module-map"; path = deepspeech_ios.modulemap; sourceTree = "<group>"; };
505B137C24961AF20007DADA /* deepspeech.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = deepspeech.h; path = ../../deepspeech.h; sourceTree = "<group>"; };
505B137E24961BA70007DADA /* DeepSpeech.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = DeepSpeech.swift; sourceTree = "<group>"; };
507CD39A24B61FA100409BBB /* libdeepspeech.so */ = {isa = PBXFileReference; lastKnownFileType = "compiled.mach-o.dylib"; path = libdeepspeech.so; sourceTree = "<group>"; };
/* End PBXFileReference section */
/* Begin PBXFrameworksBuildPhase section */
505B135E24960D550007DADA /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
507CD39B24B61FA100409BBB /* libdeepspeech.so in Frameworks */,
);
runOnlyForDeploymentPostprocessing = 0;
};
505B136724960D550007DADA /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
505B136B24960D550007DADA /* deepspeech_ios.framework in Frameworks */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXFrameworksBuildPhase section */
/* Begin PBXGroup section */
505B135724960D550007DADA = {
isa = PBXGroup;
children = (
505B136324960D550007DADA /* deepspeech_ios */,
505B136224960D550007DADA /* Products */,
505B1380249620C60007DADA /* Frameworks */,
);
sourceTree = "<group>";
};
505B136224960D550007DADA /* Products */ = {
isa = PBXGroup;
children = (
505B136124960D550007DADA /* deepspeech_ios.framework */,
505B136A24960D550007DADA /* deepspeech_iosTests.xctest */,
);
name = Products;
sourceTree = "<group>";
};
505B136324960D550007DADA /* deepspeech_ios */ = {
isa = PBXGroup;
children = (
505B137C24961AF20007DADA /* deepspeech.h */,
505B136424960D550007DADA /* deepspeech_ios.h */,
505B137E24961BA70007DADA /* DeepSpeech.swift */,
505B137B249619C90007DADA /* deepspeech_ios.modulemap */,
505B136524960D550007DADA /* Info.plist */,
);
path = deepspeech_ios;
sourceTree = "<group>";
};
505B1380249620C60007DADA /* Frameworks */ = {
isa = PBXGroup;
children = (
507CD39A24B61FA100409BBB /* libdeepspeech.so */,
);
name = Frameworks;
sourceTree = "<group>";
};
/* End PBXGroup section */
/* Begin PBXHeadersBuildPhase section */
505B135C24960D550007DADA /* Headers */ = {
isa = PBXHeadersBuildPhase;
buildActionMask = 2147483647;
files = (
505B137224960D550007DADA /* deepspeech_ios.h in Headers */,
505B137D24961AF20007DADA /* deepspeech.h in Headers */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXHeadersBuildPhase section */
/* Begin PBXNativeTarget section */
505B136024960D550007DADA /* deepspeech_ios */ = {
isa = PBXNativeTarget;
buildConfigurationList = 505B137524960D550007DADA /* Build configuration list for PBXNativeTarget "deepspeech_ios" */;
buildPhases = (
505B135C24960D550007DADA /* Headers */,
505B135D24960D550007DADA /* Sources */,
505B135E24960D550007DADA /* Frameworks */,
505B135F24960D550007DADA /* Resources */,
505B138A249628290007DADA /* Embed Libraries */,
);
buildRules = (
);
dependencies = (
);
name = deepspeech_ios;
productName = deepspeech_ios;
productReference = 505B136124960D550007DADA /* deepspeech_ios.framework */;
productType = "com.apple.product-type.framework";
};
505B136924960D550007DADA /* deepspeech_iosTests */ = {
isa = PBXNativeTarget;
buildConfigurationList = 505B137824960D550007DADA /* Build configuration list for PBXNativeTarget "deepspeech_iosTests" */;
buildPhases = (
505B136624960D550007DADA /* Sources */,
505B136724960D550007DADA /* Frameworks */,
505B136824960D550007DADA /* Resources */,
);
buildRules = (
);
dependencies = (
505B136D24960D550007DADA /* PBXTargetDependency */,
);
name = deepspeech_iosTests;
productName = deepspeech_iosTests;
productReference = 505B136A24960D550007DADA /* deepspeech_iosTests.xctest */;
productType = "com.apple.product-type.bundle.unit-test";
};
/* End PBXNativeTarget section */
/* Begin PBXProject section */
505B135824960D550007DADA /* Project object */ = {
isa = PBXProject;
attributes = {
LastSwiftUpdateCheck = 1150;
LastUpgradeCheck = 1150;
ORGANIZATIONNAME = Mozilla;
TargetAttributes = {
505B136024960D550007DADA = {
CreatedOnToolsVersion = 11.5;
LastSwiftMigration = 1150;
};
505B136924960D550007DADA = {
CreatedOnToolsVersion = 11.5;
};
};
};
buildConfigurationList = 505B135B24960D550007DADA /* Build configuration list for PBXProject "deepspeech_ios" */;
compatibilityVersion = "Xcode 9.3";
developmentRegion = en;
hasScannedForEncodings = 0;
knownRegions = (
en,
Base,
);
mainGroup = 505B135724960D550007DADA;
productRefGroup = 505B136224960D550007DADA /* Products */;
projectDirPath = "";
projectRoot = "";
targets = (
505B136024960D550007DADA /* deepspeech_ios */,
505B136924960D550007DADA /* deepspeech_iosTests */,
);
};
/* End PBXProject section */
/* Begin PBXResourcesBuildPhase section */
505B135F24960D550007DADA /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
505B136824960D550007DADA /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXResourcesBuildPhase section */
/* Begin PBXSourcesBuildPhase section */
505B135D24960D550007DADA /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
505B137F24961BA70007DADA /* DeepSpeech.swift in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
505B136624960D550007DADA /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXSourcesBuildPhase section */
/* Begin PBXTargetDependency section */
505B136D24960D550007DADA /* PBXTargetDependency */ = {
isa = PBXTargetDependency;
target = 505B136024960D550007DADA /* deepspeech_ios */;
targetProxy = 505B136C24960D550007DADA /* PBXContainerItemProxy */;
};
/* End PBXTargetDependency section */
/* Begin XCBuildConfiguration section */
505B137324960D550007DADA /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++14";
CLANG_CXX_LIBRARY = "libc++";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_ENABLE_OBJC_WEAK = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
COPY_PHASE_STRIP = NO;
CURRENT_PROJECT_VERSION = 1;
DEBUG_INFORMATION_FORMAT = dwarf;
ENABLE_STRICT_OBJC_MSGSEND = YES;
ENABLE_TESTABILITY = YES;
GCC_C_LANGUAGE_STANDARD = gnu11;
GCC_DYNAMIC_NO_PIC = NO;
GCC_NO_COMMON_BLOCKS = YES;
GCC_OPTIMIZATION_LEVEL = 0;
GCC_PREPROCESSOR_DEFINITIONS = (
"DEBUG=1",
"$(inherited)",
);
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
IPHONEOS_DEPLOYMENT_TARGET = 13.5;
MTL_ENABLE_DEBUG_INFO = INCLUDE_SOURCE;
MTL_FAST_MATH = YES;
ONLY_ACTIVE_ARCH = YES;
SDKROOT = iphoneos;
SWIFT_ACTIVE_COMPILATION_CONDITIONS = DEBUG;
SWIFT_OPTIMIZATION_LEVEL = "-Onone";
VERSIONING_SYSTEM = "apple-generic";
VERSION_INFO_PREFIX = "";
};
name = Debug;
};
505B137424960D550007DADA /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++14";
CLANG_CXX_LIBRARY = "libc++";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_ENABLE_OBJC_WEAK = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
COPY_PHASE_STRIP = NO;
CURRENT_PROJECT_VERSION = 1;
DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
ENABLE_NS_ASSERTIONS = NO;
ENABLE_STRICT_OBJC_MSGSEND = YES;
GCC_C_LANGUAGE_STANDARD = gnu11;
GCC_NO_COMMON_BLOCKS = YES;
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
IPHONEOS_DEPLOYMENT_TARGET = 13.5;
MTL_ENABLE_DEBUG_INFO = NO;
MTL_FAST_MATH = YES;
SDKROOT = iphoneos;
SWIFT_COMPILATION_MODE = wholemodule;
SWIFT_OPTIMIZATION_LEVEL = "-O";
VALIDATE_PRODUCT = YES;
VERSIONING_SYSTEM = "apple-generic";
VERSION_INFO_PREFIX = "";
};
name = Release;
};
505B137624960D550007DADA /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
CLANG_ENABLE_MODULES = YES;
CODE_SIGN_IDENTITY = "Apple Development";
"CODE_SIGN_IDENTITY[sdk=macosx*]" = "Apple Development";
CODE_SIGN_STYLE = Automatic;
DEFINES_MODULE = YES;
DEVELOPMENT_TEAM = AWCG9S27P7;
DYLIB_COMPATIBILITY_VERSION = 1;
DYLIB_CURRENT_VERSION = 1;
DYLIB_INSTALL_NAME_BASE = "@rpath";
INFOPLIST_FILE = deepspeech_ios/Info.plist;
INSTALL_PATH = "$(LOCAL_LIBRARY_DIR)/Frameworks";
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
LIBRARY_SEARCH_PATHS = (
"$(inherited)",
"$(PROJECT_DIR)/deepspeech_ios",
"$(PROJECT_DIR)",
);
MODULEMAP_FILE = deepspeech_ios/deepspeech_ios.modulemap;
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios";
PRODUCT_NAME = "$(TARGET_NAME:c99extidentifier)";
SKIP_INSTALL = YES;
SWIFT_OPTIMIZATION_LEVEL = "-Onone";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Debug;
};
505B137724960D550007DADA /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
CLANG_ENABLE_MODULES = YES;
CODE_SIGN_IDENTITY = "Apple Development";
"CODE_SIGN_IDENTITY[sdk=macosx*]" = "Apple Development";
CODE_SIGN_STYLE = Automatic;
DEFINES_MODULE = YES;
DEVELOPMENT_TEAM = AWCG9S27P7;
DYLIB_COMPATIBILITY_VERSION = 1;
DYLIB_CURRENT_VERSION = 1;
DYLIB_INSTALL_NAME_BASE = "@rpath";
INFOPLIST_FILE = deepspeech_ios/Info.plist;
INSTALL_PATH = "$(LOCAL_LIBRARY_DIR)/Frameworks";
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
LIBRARY_SEARCH_PATHS = (
"$(inherited)",
"$(PROJECT_DIR)/deepspeech_ios",
"$(PROJECT_DIR)",
);
MODULEMAP_FILE = deepspeech_ios/deepspeech_ios.modulemap;
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios";
PRODUCT_NAME = "$(TARGET_NAME:c99extidentifier)";
SKIP_INSTALL = YES;
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Release;
};
505B137924960D550007DADA /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_EMBED_SWIFT_STANDARD_LIBRARIES = YES;
CODE_SIGN_STYLE = Automatic;
INFOPLIST_FILE = deepspeech_iosTests/Info.plist;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-iosTests";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Debug;
};
505B137A24960D550007DADA /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_EMBED_SWIFT_STANDARD_LIBRARIES = YES;
CODE_SIGN_STYLE = Automatic;
INFOPLIST_FILE = deepspeech_iosTests/Info.plist;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-iosTests";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Release;
};
/* End XCBuildConfiguration section */
/* Begin XCConfigurationList section */
505B135B24960D550007DADA /* Build configuration list for PBXProject "deepspeech_ios" */ = {
isa = XCConfigurationList;
buildConfigurations = (
505B137324960D550007DADA /* Debug */,
505B137424960D550007DADA /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
505B137524960D550007DADA /* Build configuration list for PBXNativeTarget "deepspeech_ios" */ = {
isa = XCConfigurationList;
buildConfigurations = (
505B137624960D550007DADA /* Debug */,
505B137724960D550007DADA /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
505B137824960D550007DADA /* Build configuration list for PBXNativeTarget "deepspeech_iosTests" */ = {
isa = XCConfigurationList;
buildConfigurations = (
505B137924960D550007DADA /* Debug */,
505B137A24960D550007DADA /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
/* End XCConfigurationList section */
};
rootObject = 505B135824960D550007DADA /* Project object */;
}

7
native_client/swift/deepspeech_ios.xcodeproj/project.xcworkspace/contents.xcworkspacedata generated Normal file
View File

@ -0,0 +1,7 @@
<?xml version="1.0" encoding="UTF-8"?>
<Workspace
version = "1.0">
<FileRef
location = "self:deepspeech_ios.xcodeproj">
</FileRef>
</Workspace>

8
native_client/swift/deepspeech_ios.xcodeproj/project.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist Normal file
View File

@ -0,0 +1,8 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>IDEDidComputeMac32BitWarning</key>
<true/>
</dict>
</plist>

77
native_client/swift/deepspeech_ios.xcodeproj/xcshareddata/xcschemes/deepspeech_ios.xcscheme Normal file
View File

@ -0,0 +1,77 @@
<?xml version="1.0" encoding="UTF-8"?>
<Scheme
LastUpgradeVersion = "1150"
version = "1.3">
<BuildAction
parallelizeBuildables = "YES"
buildImplicitDependencies = "YES">
<BuildActionEntries>
<BuildActionEntry
buildForTesting = "YES"
buildForRunning = "YES"
buildForProfiling = "YES"
buildForArchiving = "YES"
buildForAnalyzing = "YES">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "505B136024960D550007DADA"
BuildableName = "deepspeech_ios.framework"
BlueprintName = "deepspeech_ios"
ReferencedContainer = "container:deepspeech_ios.xcodeproj">
</BuildableReference>
</BuildActionEntry>
</BuildActionEntries>
</BuildAction>
<TestAction
buildConfiguration = "Debug"
selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
shouldUseLaunchSchemeArgsEnv = "YES">
<Testables>
<TestableReference
skipped = "NO">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "505B136924960D550007DADA"
BuildableName = "deepspeech_iosTests.xctest"
BlueprintName = "deepspeech_iosTests"
ReferencedContainer = "container:deepspeech_ios.xcodeproj">
</BuildableReference>
</TestableReference>
</Testables>
</TestAction>
<LaunchAction
buildConfiguration = "Debug"
selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
launchStyle = "0"
useCustomWorkingDirectory = "NO"
ignoresPersistentStateOnLaunch = "NO"
debugDocumentVersioning = "YES"
debugServiceExtension = "internal"
allowLocationSimulation = "YES">
</LaunchAction>
<ProfileAction
buildConfiguration = "Release"
shouldUseLaunchSchemeArgsEnv = "YES"
savedToolIdentifier = ""
useCustomWorkingDirectory = "NO"
debugDocumentVersioning = "YES">
<MacroExpansion>
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "505B136024960D550007DADA"
BuildableName = "deepspeech_ios.framework"
BlueprintName = "deepspeech_ios"
ReferencedContainer = "container:deepspeech_ios.xcodeproj">
</BuildableReference>
</MacroExpansion>
</ProfileAction>
<AnalyzeAction
buildConfiguration = "Debug">
</AnalyzeAction>
<ArchiveAction
buildConfiguration = "Release"
revealArchiveInOrganizer = "YES">
</ArchiveAction>
</Scheme>

10
native_client/swift/deepspeech_ios.xcworkspace/contents.xcworkspacedata generated Normal file
View File

@ -0,0 +1,10 @@
<?xml version="1.0" encoding="UTF-8"?>
<Workspace
version = "1.0">
<FileRef
location = "group:deepspeech_ios.xcodeproj">
</FileRef>
<FileRef
location = "group:deepspeech_ios_test.xcodeproj">
</FileRef>
</Workspace>

8
native_client/swift/deepspeech_ios.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist Normal file
View File

@ -0,0 +1,8 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>IDEDidComputeMac32BitWarning</key>
<true/>
</dict>
</plist>

8
native_client/swift/deepspeech_ios.xcworkspace/xcshareddata/WorkspaceSettings.xcsettings Normal file
View File

@ -0,0 +1,8 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>PreviewsEnabled</key>
<false/>
</dict>
</plist>

454
native_client/swift/deepspeech_ios/DeepSpeech.swift Normal file
View File

@ -0,0 +1,454 @@
//
// DeepSpeech.swift
// deepspeech_ios
//
// Created by Reuben Morais on 14.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
import deepspeech_ios.libdeepspeech_Private
public enum DeepSpeechError: Error {
// Should be kept in sync with deepspeech.h
case noModel(errorCode: Int32)
case invalidAlphabet(errorCode: Int32)
case invalidShape(errorCode: Int32)
case invalidScorer(errorCode: Int32)
case modelIncompatible(errorCode: Int32)
case scorerNotEnabled(errorCode: Int32)
case scorerUnreadable(errorCode: Int32)
case scorerInvalidLm(errorCode: Int32)
case scorerNoTrie(errorCode: Int32)
case scorerInvalidTrie(errorCode: Int32)
case scorerVersionMismatch(errorCode: Int32)
case failInitMmap(errorCode: Int32)
case failInitSess(errorCode: Int32)
case failInterpreter(errorCode: Int32)
case failRunSess(errorCode: Int32)
case failCreateStream(errorCode: Int32)
case failReadProtobuf(errorCode: Int32)
case failCreateSess(errorCode: Int32)
case failCreateModel(errorCode: Int32)
// Additional case for invalid error codes, should never happen unless the
// user has mixed header and binary versions.
case invalidErrorCode(errorCode: Int32)
}
extension DeepSpeechError : LocalizedError {
public var errorDescription: String? {
switch self {
case .noModel(let errorCode),
.invalidAlphabet(let errorCode),
.invalidShape(let errorCode),
.invalidScorer(let errorCode),
.modelIncompatible(let errorCode),
.scorerNotEnabled(let errorCode),
.scorerUnreadable(let errorCode),
.scorerInvalidLm(let errorCode),
.scorerNoTrie(let errorCode),
.scorerInvalidTrie(let errorCode),
.scorerVersionMismatch(let errorCode),
.failInitMmap(let errorCode),
.failInitSess(let errorCode),
.failInterpreter(let errorCode),
.failRunSess(let errorCode),
.failCreateStream(let errorCode),
.failReadProtobuf(let errorCode),
.failCreateSess(let errorCode),
.failCreateModel(let errorCode),
.invalidErrorCode(let errorCode):
let result = DS_ErrorCodeToErrorMessage(errorCode)
defer { DS_FreeString(result) }
return String(cString: result!)
}
}
}
private func errorCodeToEnum(errorCode: Int32) -> DeepSpeechError {
switch Int(errorCode) {
case Int(DS_ERR_NO_MODEL.rawValue):
return DeepSpeechError.noModel(errorCode: errorCode)
case Int(DS_ERR_INVALID_ALPHABET.rawValue):
return DeepSpeechError.invalidAlphabet(errorCode: errorCode)
case Int(DS_ERR_INVALID_SHAPE.rawValue):
return DeepSpeechError.invalidShape(errorCode: errorCode)
case Int(DS_ERR_INVALID_SCORER.rawValue):
return DeepSpeechError.invalidScorer(errorCode: errorCode)
case Int(DS_ERR_MODEL_INCOMPATIBLE.rawValue):
return DeepSpeechError.modelIncompatible(errorCode: errorCode)
case Int(DS_ERR_SCORER_NOT_ENABLED.rawValue):
return DeepSpeechError.scorerNotEnabled(errorCode: errorCode)
case Int(DS_ERR_SCORER_UNREADABLE.rawValue):
return DeepSpeechError.scorerUnreadable(errorCode: errorCode)
case Int(DS_ERR_SCORER_INVALID_LM.rawValue):
return DeepSpeechError.scorerInvalidLm(errorCode: errorCode)
case Int(DS_ERR_SCORER_NO_TRIE.rawValue):
return DeepSpeechError.scorerNoTrie(errorCode: errorCode)
case Int(DS_ERR_SCORER_INVALID_TRIE.rawValue):
return DeepSpeechError.scorerInvalidTrie(errorCode: errorCode)
case Int(DS_ERR_SCORER_VERSION_MISMATCH.rawValue):
return DeepSpeechError.scorerVersionMismatch(errorCode: errorCode)
case Int(DS_ERR_FAIL_INIT_MMAP.rawValue):
return DeepSpeechError.failInitMmap(errorCode: errorCode)
case Int(DS_ERR_FAIL_INIT_SESS.rawValue):
return DeepSpeechError.failInitSess(errorCode: errorCode)
case Int(DS_ERR_FAIL_INTERPRETER.rawValue):
return DeepSpeechError.failInterpreter(errorCode: errorCode)
case Int(DS_ERR_FAIL_RUN_SESS.rawValue):
return DeepSpeechError.failRunSess(errorCode: errorCode)
case Int(DS_ERR_FAIL_CREATE_STREAM.rawValue):
return DeepSpeechError.failCreateStream(errorCode: errorCode)
case Int(DS_ERR_FAIL_READ_PROTOBUF.rawValue):
return DeepSpeechError.failReadProtobuf(errorCode: errorCode)
case Int(DS_ERR_FAIL_CREATE_SESS.rawValue):
return DeepSpeechError.failCreateSess(errorCode: errorCode)
case Int(DS_ERR_FAIL_CREATE_MODEL.rawValue):
return DeepSpeechError.failCreateModel(errorCode: errorCode)
default:
return DeepSpeechError.invalidErrorCode(errorCode: errorCode)
}
}
private func evaluateErrorCode(errorCode: Int32) throws {
if errorCode != Int32(DS_ERR_OK.rawValue) {
throw errorCodeToEnum(errorCode: errorCode)
}
}
/// Stores text of an individual token, along with its timing information
public struct DeepSpeechTokenMetadata {
/// The text corresponding to this token
let text: String
/// Position of the token in units of 20ms
let timestep: Int
/// Position of the token in seconds
let startTime: Float
internal init(fromInternal: TokenMetadata) {
text = String(cString: fromInternal.text)
timestep = Int(fromInternal.timestep)
startTime = fromInternal.start_time
}
}
/** A single transcript computed by the model, including a confidence value and
the metadata for its constituent tokens
*/
public struct DeepSpeechCandidateTranscript {
/// Array of DeepSpeechTokenMetadata objects
private(set) var tokens: [DeepSpeechTokenMetadata] = []
/** Approximated confidence value for this transcript. This corresponds to
both acoustic model and language model scores that contributed to the
creation of this transcript.
*/
let confidence: Double
internal init(fromInternal: CandidateTranscript) {
let tokensBuffer = UnsafeBufferPointer<TokenMetadata>(start: fromInternal.tokens, count: Int(fromInternal.num_tokens))
for tok in tokensBuffer {
tokens.append(DeepSpeechTokenMetadata(fromInternal: tok))
}
confidence = fromInternal.confidence
}
}
/// An array of DeepSpeechCandidateTranscript objects computed by the model
public struct DeepSpeechMetadata {
/// Array of DeepSpeechCandidateTranscript objects
private(set) var transcripts: [DeepSpeechCandidateTranscript] = []
internal init(fromInternal: UnsafeMutablePointer<Metadata>) {
let md = fromInternal.pointee
let transcriptsBuffer = UnsafeBufferPointer<CandidateTranscript>(
start: md.transcripts,
count: Int(md.num_transcripts))
for tr in transcriptsBuffer {
transcripts.append(DeepSpeechCandidateTranscript(fromInternal: tr))
}
}
}
public class DeepSpeechStream {
private var streamCtx: OpaquePointer!
internal init(streamContext: OpaquePointer) {
streamCtx = streamContext
}
deinit {
if streamCtx != nil {
DS_FreeStream(streamCtx)
streamCtx = nil
}
}
/** Feed audio samples to an ongoing streaming inference.
- Parameter buffer: A 16-bit, mono raw audio signal at the appropriate
sample rate (matching what the model was trained on).
- Precondition: `finishStream()` has not been called on this stream.
*/
public func feedAudioContent(buffer: Array<Int16>) {
precondition(streamCtx != nil, "calling method on invalidated Stream")
buffer.withUnsafeBufferPointer { unsafeBufferPointer in
feedAudioContent(buffer: unsafeBufferPointer)
}
}
/** Feed audio samples to an ongoing streaming inference.
- Parameter buffer: A 16-bit, mono raw audio signal at the appropriate
sample rate (matching what the model was trained on).
- Precondition: `finishStream()` has not been called on this stream.
*/
public func feedAudioContent(buffer: UnsafeBufferPointer<Int16>) {
precondition(streamCtx != nil, "calling method on invalidated Stream")
DS_FeedAudioContent(streamCtx, buffer.baseAddress, UInt32(buffer.count))
}
/** Compute the intermediate decoding of an ongoing streaming inference.
- Precondition: `finishStream()` has not been called on this stream.
- Returns: The STT intermediate result.
*/
public func intermediateDecode() -> String {
precondition(streamCtx != nil, "calling method on invalidated Stream")
let result = DS_IntermediateDecode(streamCtx)
defer { DS_FreeString(result) }
return String(cString: result!)
}
/** Compute the intermediate decoding of an ongoing streaming inference,
return results including metadata.
- Parameter numResults: The number of candidate transcripts to return.
- Precondition: `finishStream()` has not been called on this stream.
- Returns: Metadata struct containing multiple CandidateTranscript structs.
Each transcript has per-token metadata including timing information.
*/
public func intermediateDecodeWithMetadata(numResults: Int) -> DeepSpeechMetadata {
precondition(streamCtx != nil, "calling method on invalidated Stream")
let result = DS_IntermediateDecodeWithMetadata(streamCtx, UInt32(numResults))!
defer { DS_FreeMetadata(result) }
return DeepSpeechMetadata(fromInternal: result)
}
/** Compute the final decoding of an ongoing streaming inference and return
the result. Signals the end of an ongoing streaming inference.
- Precondition: `finishStream()` has not been called on this stream.
- Returns: The STT result.
- Postcondition: This method will invalidate this streaming context.
*/
public func finishStream() -> String {
precondition(streamCtx != nil, "calling method on invalidated Stream")
let result = DS_FinishStream(streamCtx)
defer {
DS_FreeString(result)
streamCtx = nil
}
return String(cString: result!)
}
/** Compute the final decoding of an ongoing streaming inference and return
results including metadata. Signals the end of an ongoing streaming
inference.
- Parameter numResults: The number of candidate transcripts to return.
- Precondition: `finishStream()` has not been called on this stream.
- Returns: Metadata struct containing multiple CandidateTranscript structs.
Each transcript has per-token metadata including timing information.
- Postcondition: This method will invalidate this streaming context.
*/
public func finishStreamWithMetadata(numResults: Int) -> DeepSpeechMetadata {
precondition(streamCtx != nil, "calling method on invalidated Stream")
let result = DS_FinishStreamWithMetadata(streamCtx, UInt32(numResults))!
defer { DS_FreeMetadata(result) }
return DeepSpeechMetadata(fromInternal: result)
}
}
/// An object providing an interface to a trained DeepSpeech model.
public class DeepSpeechModel {
private var modelCtx: OpaquePointer!
/**
- Parameter modelPath: The path to the model file.
- Throws: `DeepSpeechError` on failure.
*/
public init(modelPath: String) throws {
let err = DS_CreateModel(modelPath, &modelCtx)
try evaluateErrorCode(errorCode: err)
}
deinit {
DS_FreeModel(modelCtx)
modelCtx = nil
}
/** Get beam width value used by the model. If {@link DS_SetModelBeamWidth}
was not called before, will return the default value loaded from the
model file.
- Returns: Beam width value used by the model.
*/
public func getBeamWidth() -> Int {
return Int(DS_GetModelBeamWidth(modelCtx))
}
/** Set beam width value used by the model.
- Parameter beamWidth: The beam width used by the model. A larger beam
width value generates better results at the cost
of decoding time.
- Throws: `DeepSpeechError` on failure.
*/
public func setBeamWidth(beamWidth: Int) throws {
let err = DS_SetModelBeamWidth(modelCtx, UInt32(beamWidth))
try evaluateErrorCode(errorCode: err)
}
// The sample rate expected by the model.
public var sampleRate: Int {
get {
return Int(DS_GetModelSampleRate(modelCtx))
}
}
/** Enable decoding using an external scorer.
- Parameter scorerPath: The path to the external scorer file.
- Throws: `DeepSpeechError` on failure.
*/
public func enableExternalScorer(scorerPath: String) throws {
let err = DS_EnableExternalScorer(modelCtx, scorerPath)
try evaluateErrorCode(errorCode: err)
}
/** Disable decoding using an external scorer.
- Throws: `DeepSpeechError` on failure.
*/
public func disableExternalScorer() throws {
let err = DS_DisableExternalScorer(modelCtx)
try evaluateErrorCode(errorCode: err)
}
/** Set hyperparameters alpha and beta of the external scorer.
- Parameter alpha: The alpha hyperparameter of the decoder. Language model weight.
- Parameter beta: The beta hyperparameter of the decoder. Word insertion weight.
- Throws: `DeepSpeechError` on failure.
*/
public func setScorerAlphaBeta(alpha: Float, beta: Float) throws {
let err = DS_SetScorerAlphaBeta(modelCtx, alpha, beta)
try evaluateErrorCode(errorCode: err)
}
/** Use the DeepSpeech model to convert speech to text.
- Parameter buffer: A 16-bit, mono raw audio signal at the appropriate
sample rate (matching what the model was trained on).
- Returns: The STT result.
*/
public func speechToText(buffer: Array<Int16>) -> String {
return buffer.withUnsafeBufferPointer { unsafeBufferPointer -> String in
return speechToText(buffer: unsafeBufferPointer)
}
}
/** Use the DeepSpeech model to convert speech to text.
- Parameter buffer: A 16-bit, mono raw audio signal at the appropriate
sample rate (matching what the model was trained on).
- Returns: The STT result.
*/
public func speechToText(buffer: UnsafeBufferPointer<Int16>) -> String {
let result = DS_SpeechToText(modelCtx, buffer.baseAddress, UInt32(buffer.count))
defer { DS_FreeString(result) }
return String(cString: result!)
}
/** Use the DeepSpeech model to convert speech to text and output results
including metadata.
- Parameter buffer: A 16-bit, mono raw audio signal at the appropriate
sample rate (matching what the model was trained on).
- Parameter numResults: The maximum number of DeepSpeechCandidateTranscript
structs to return. Returned value might be smaller than this.
- Returns: Metadata struct containing multiple CandidateTranscript structs.
Each transcript has per-token metadata including timing information.
*/
public func speechToTextWithMetadata(buffer: Array<Int16>, numResults: Int) -> DeepSpeechMetadata {
return buffer.withUnsafeBufferPointer { unsafeBufferPointer -> DeepSpeechMetadata in
return speechToTextWithMetadata(buffer: unsafeBufferPointer, numResults: numResults)
}
}
/** Use the DeepSpeech model to convert speech to text and output results
including metadata.
- Parameter buffer: A 16-bit, mono raw audio signal at the appropriate
sample rate (matching what the model was trained on).
- Parameter numResults: The maximum number of DeepSpeechCandidateTranscript
structs to return. Returned value might be smaller than this.
- Returns: Metadata struct containing multiple CandidateTranscript structs.
Each transcript has per-token metadata including timing information.
*/
public func speechToTextWithMetadata(buffer: UnsafeBufferPointer<Int16>, numResults: Int) -> DeepSpeechMetadata {
let result = DS_SpeechToTextWithMetadata(
modelCtx,
buffer.baseAddress,
UInt32(buffer.count),
UInt32(numResults))!
defer { DS_FreeMetadata(result) }
return DeepSpeechMetadata(fromInternal: result)
}
/** Create a new streaming inference state.
- Returns: DeepSpeechStream object representing the streaming state.
- Throws: `DeepSpeechError` on failure.
*/
public func createStream() throws -> DeepSpeechStream {
var streamContext: OpaquePointer!
let err = DS_CreateStream(modelCtx, &streamContext)
try evaluateErrorCode(errorCode: err)
return DeepSpeechStream(streamContext: streamContext)
}
}
public func DeepSpeechVersion() -> String {
let result = DS_Version()
defer { DS_FreeString(result) }
return String(cString: result!)
}

22
native_client/swift/deepspeech_ios/Info.plist Normal file
View File

@ -0,0 +1,22 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>$(DEVELOPMENT_LANGUAGE)</string>
<key>CFBundleExecutable</key>
<string>$(EXECUTABLE_NAME)</string>
<key>CFBundleIdentifier</key>
<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>$(PRODUCT_NAME)</string>
<key>CFBundlePackageType</key>
<string>$(PRODUCT_BUNDLE_PACKAGE_TYPE)</string>
<key>CFBundleShortVersionString</key>
<string>1.0</string>
<key>CFBundleVersion</key>
<string>$(CURRENT_PROJECT_VERSION)</string>
</dict>
</plist>

13
native_client/swift/deepspeech_ios/deepspeech_ios.h Normal file
View File

@ -0,0 +1,13 @@
//
// deepspeech_ios.h
// deepspeech_ios
//
// Created by Reuben Morais on 14.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
#import <Foundation/Foundation.h>
// In this header, you should import all the public headers of your framework using statements like #import <deepspeech_ios/PublicHeader.h>

12
native_client/swift/deepspeech_ios/deepspeech_ios.modulemap Normal file
View File

@ -0,0 +1,12 @@
framework module deepspeech_ios {
umbrella header "deepspeech_ios.h"
export *
module * { export * }
explicit module libdeepspeech_Private {
header "deepspeech.h"
export *
link "deepspeech"
}
}

637
native_client/swift/deepspeech_ios_test.xcodeproj/project.pbxproj Normal file
View File

@ -0,0 +1,637 @@
// !$*UTF8*$!
{
archiveVersion = 1;
classes = {
};
objectVersion = 50;
objects = {
/* Begin PBXBuildFile section */
507CD3A124B61FE400409BBB /* deepspeech_ios.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 507CD3A024B61FE400409BBB /* deepspeech_ios.framework */; };
507CD3A324B61FEB00409BBB /* libdeepspeech.so in Frameworks */ = {isa = PBXBuildFile; fileRef = 507CD3A224B61FEA00409BBB /* libdeepspeech.so */; };
507CD3A424B61FFC00409BBB /* libdeepspeech.so in Embed Frameworks */ = {isa = PBXBuildFile; fileRef = 507CD3A224B61FEA00409BBB /* libdeepspeech.so */; settings = {ATTRIBUTES = (CodeSignOnCopy, ); }; };
50F787F32497683900D52237 /* AppDelegate.swift in Sources */ = {isa = PBXBuildFile; fileRef = 50F787F22497683900D52237 /* AppDelegate.swift */; };
50F787F52497683900D52237 /* SceneDelegate.swift in Sources */ = {isa = PBXBuildFile; fileRef = 50F787F42497683900D52237 /* SceneDelegate.swift */; };
50F787F72497683900D52237 /* ContentView.swift in Sources */ = {isa = PBXBuildFile; fileRef = 50F787F62497683900D52237 /* ContentView.swift */; };
50F787F92497683A00D52237 /* Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 50F787F82497683A00D52237 /* Assets.xcassets */; };
50F787FC2497683A00D52237 /* Preview Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 50F787FB2497683A00D52237 /* Preview Assets.xcassets */; };
50F787FF2497683A00D52237 /* LaunchScreen.storyboard in Resources */ = {isa = PBXBuildFile; fileRef = 50F787FD2497683A00D52237 /* LaunchScreen.storyboard */; };
50F7880A2497683A00D52237 /* deepspeech_ios_testTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = 50F788092497683A00D52237 /* deepspeech_ios_testTests.swift */; };
50F788152497683A00D52237 /* deepspeech_ios_testUITests.swift in Sources */ = {isa = PBXBuildFile; fileRef = 50F788142497683A00D52237 /* deepspeech_ios_testUITests.swift */; };
/* End PBXBuildFile section */
/* Begin PBXContainerItemProxy section */
50F788062497683A00D52237 /* PBXContainerItemProxy */ = {
isa = PBXContainerItemProxy;
containerPortal = 50F787E72497683900D52237 /* Project object */;
proxyType = 1;
remoteGlobalIDString = 50F787EE2497683900D52237;
remoteInfo = deepspeech_ios_test;
};
50F788112497683A00D52237 /* PBXContainerItemProxy */ = {
isa = PBXContainerItemProxy;
containerPortal = 50F787E72497683900D52237 /* Project object */;
proxyType = 1;
remoteGlobalIDString = 50F787EE2497683900D52237;
remoteInfo = deepspeech_ios_test;
};
/* End PBXContainerItemProxy section */
/* Begin PBXCopyFilesBuildPhase section */
50F2B10E2498EB59007CD876 /* Embed Frameworks */ = {
isa = PBXCopyFilesBuildPhase;
buildActionMask = 2147483647;
dstPath = "";
dstSubfolderSpec = 10;
files = (
507CD3A424B61FFC00409BBB /* libdeepspeech.so in Embed Frameworks */,
);
name = "Embed Frameworks";
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXCopyFilesBuildPhase section */
/* Begin PBXFileReference section */
507CD3A024B61FE400409BBB /* deepspeech_ios.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; path = deepspeech_ios.framework; sourceTree = BUILT_PRODUCTS_DIR; };
507CD3A224B61FEA00409BBB /* libdeepspeech.so */ = {isa = PBXFileReference; lastKnownFileType = "compiled.mach-o.dylib"; name = libdeepspeech.so; path = libdeepspeech.so; sourceTree = "<group>"; };
50F787EF2497683900D52237 /* deepspeech_ios_test.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = deepspeech_ios_test.app; sourceTree = BUILT_PRODUCTS_DIR; };
50F787F22497683900D52237 /* AppDelegate.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = AppDelegate.swift; sourceTree = "<group>"; };
50F787F42497683900D52237 /* SceneDelegate.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = SceneDelegate.swift; sourceTree = "<group>"; };
50F787F62497683900D52237 /* ContentView.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = ContentView.swift; sourceTree = "<group>"; };
50F787F82497683A00D52237 /* Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = Assets.xcassets; sourceTree = "<group>"; };
50F787FB2497683A00D52237 /* Preview Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = "Preview Assets.xcassets"; sourceTree = "<group>"; };
50F787FE2497683A00D52237 /* Base */ = {isa = PBXFileReference; lastKnownFileType = file.storyboard; name = Base; path = Base.lproj/LaunchScreen.storyboard; sourceTree = "<group>"; };
50F788002497683A00D52237 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
50F788052497683A00D52237 /* deepspeech_ios_testTests.xctest */ = {isa = PBXFileReference; explicitFileType = wrapper.cfbundle; includeInIndex = 0; path = deepspeech_ios_testTests.xctest; sourceTree = BUILT_PRODUCTS_DIR; };
50F788092497683A00D52237 /* deepspeech_ios_testTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = deepspeech_ios_testTests.swift; sourceTree = "<group>"; };
50F7880B2497683A00D52237 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
50F788102497683A00D52237 /* deepspeech_ios_testUITests.xctest */ = {isa = PBXFileReference; explicitFileType = wrapper.cfbundle; includeInIndex = 0; path = deepspeech_ios_testUITests.xctest; sourceTree = BUILT_PRODUCTS_DIR; };
50F788142497683A00D52237 /* deepspeech_ios_testUITests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = deepspeech_ios_testUITests.swift; sourceTree = "<group>"; };
50F788162497683A00D52237 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
/* End PBXFileReference section */
/* Begin PBXFrameworksBuildPhase section */
50F787EC2497683900D52237 /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
507CD3A324B61FEB00409BBB /* libdeepspeech.so in Frameworks */,
507CD3A124B61FE400409BBB /* deepspeech_ios.framework in Frameworks */,
);
runOnlyForDeploymentPostprocessing = 0;
};
50F788022497683A00D52237 /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
50F7880D2497683A00D52237 /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXFrameworksBuildPhase section */
/* Begin PBXGroup section */
50F2B0FC2498D6C7007CD876 /* Frameworks */ = {
isa = PBXGroup;
children = (
507CD3A224B61FEA00409BBB /* libdeepspeech.so */,
507CD3A024B61FE400409BBB /* deepspeech_ios.framework */,
);
name = Frameworks;
sourceTree = "<group>";
};
50F787E62497683900D52237 = {
isa = PBXGroup;
children = (
50F787F12497683900D52237 /* deepspeech_ios_test */,
50F788082497683A00D52237 /* deepspeech_ios_testTests */,
50F788132497683A00D52237 /* deepspeech_ios_testUITests */,
50F787F02497683900D52237 /* Products */,
50F2B0FC2498D6C7007CD876 /* Frameworks */,
);
sourceTree = "<group>";
};
50F787F02497683900D52237 /* Products */ = {
isa = PBXGroup;
children = (
50F787EF2497683900D52237 /* deepspeech_ios_test.app */,
50F788052497683A00D52237 /* deepspeech_ios_testTests.xctest */,
50F788102497683A00D52237 /* deepspeech_ios_testUITests.xctest */,
);
name = Products;
sourceTree = "<group>";
};
50F787F12497683900D52237 /* deepspeech_ios_test */ = {
isa = PBXGroup;
children = (
50F787F22497683900D52237 /* AppDelegate.swift */,
50F787F42497683900D52237 /* SceneDelegate.swift */,
50F787F62497683900D52237 /* ContentView.swift */,
50F787F82497683A00D52237 /* Assets.xcassets */,
50F787FD2497683A00D52237 /* LaunchScreen.storyboard */,
50F788002497683A00D52237 /* Info.plist */,
50F787FA2497683A00D52237 /* Preview Content */,
);
path = deepspeech_ios_test;
sourceTree = "<group>";
};
50F787FA2497683A00D52237 /* Preview Content */ = {
isa = PBXGroup;
children = (
50F787FB2497683A00D52237 /* Preview Assets.xcassets */,
);
path = "Preview Content";
sourceTree = "<group>";
};
50F788082497683A00D52237 /* deepspeech_ios_testTests */ = {
isa = PBXGroup;
children = (
50F788092497683A00D52237 /* deepspeech_ios_testTests.swift */,
50F7880B2497683A00D52237 /* Info.plist */,
);
path = deepspeech_ios_testTests;
sourceTree = "<group>";
};
50F788132497683A00D52237 /* deepspeech_ios_testUITests */ = {
isa = PBXGroup;
children = (
50F788142497683A00D52237 /* deepspeech_ios_testUITests.swift */,
50F788162497683A00D52237 /* Info.plist */,
);
path = deepspeech_ios_testUITests;
sourceTree = "<group>";
};
/* End PBXGroup section */
/* Begin PBXNativeTarget section */
50F787EE2497683900D52237 /* deepspeech_ios_test */ = {
isa = PBXNativeTarget;
buildConfigurationList = 50F788192497683A00D52237 /* Build configuration list for PBXNativeTarget "deepspeech_ios_test" */;
buildPhases = (
50F787EB2497683900D52237 /* Sources */,
50F787EC2497683900D52237 /* Frameworks */,
50F787ED2497683900D52237 /* Resources */,
50F2B10E2498EB59007CD876 /* Embed Frameworks */,
);
buildRules = (
);
dependencies = (
);
name = deepspeech_ios_test;
productName = deepspeech_ios_test;
productReference = 50F787EF2497683900D52237 /* deepspeech_ios_test.app */;
productType = "com.apple.product-type.application";
};
50F788042497683A00D52237 /* deepspeech_ios_testTests */ = {
isa = PBXNativeTarget;
buildConfigurationList = 50F7881C2497683A00D52237 /* Build configuration list for PBXNativeTarget "deepspeech_ios_testTests" */;
buildPhases = (
50F788012497683A00D52237 /* Sources */,
50F788022497683A00D52237 /* Frameworks */,
50F788032497683A00D52237 /* Resources */,
);
buildRules = (
);
dependencies = (
50F788072497683A00D52237 /* PBXTargetDependency */,
);
name = deepspeech_ios_testTests;
productName = deepspeech_ios_testTests;
productReference = 50F788052497683A00D52237 /* deepspeech_ios_testTests.xctest */;
productType = "com.apple.product-type.bundle.unit-test";
};
50F7880F2497683A00D52237 /* deepspeech_ios_testUITests */ = {
isa = PBXNativeTarget;
buildConfigurationList = 50F7881F2497683A00D52237 /* Build configuration list for PBXNativeTarget "deepspeech_ios_testUITests" */;
buildPhases = (
50F7880C2497683A00D52237 /* Sources */,
50F7880D2497683A00D52237 /* Frameworks */,
50F7880E2497683A00D52237 /* Resources */,
);
buildRules = (
);
dependencies = (
50F788122497683A00D52237 /* PBXTargetDependency */,
);
name = deepspeech_ios_testUITests;
productName = deepspeech_ios_testUITests;
productReference = 50F788102497683A00D52237 /* deepspeech_ios_testUITests.xctest */;
productType = "com.apple.product-type.bundle.ui-testing";
};
/* End PBXNativeTarget section */
/* Begin PBXProject section */
50F787E72497683900D52237 /* Project object */ = {
isa = PBXProject;
attributes = {
LastSwiftUpdateCheck = 1150;
LastUpgradeCheck = 1150;
ORGANIZATIONNAME = Mozilla;
TargetAttributes = {
50F787EE2497683900D52237 = {
CreatedOnToolsVersion = 11.5;
};
50F788042497683A00D52237 = {
CreatedOnToolsVersion = 11.5;
TestTargetID = 50F787EE2497683900D52237;
};
50F7880F2497683A00D52237 = {
CreatedOnToolsVersion = 11.5;
TestTargetID = 50F787EE2497683900D52237;
};
};
};
buildConfigurationList = 50F787EA2497683900D52237 /* Build configuration list for PBXProject "deepspeech_ios_test" */;
compatibilityVersion = "Xcode 9.3";
developmentRegion = en;
hasScannedForEncodings = 0;
knownRegions = (
en,
Base,
);
mainGroup = 50F787E62497683900D52237;
productRefGroup = 50F787F02497683900D52237 /* Products */;
projectDirPath = "";
projectRoot = "";
targets = (
50F787EE2497683900D52237 /* deepspeech_ios_test */,
50F788042497683A00D52237 /* deepspeech_ios_testTests */,
50F7880F2497683A00D52237 /* deepspeech_ios_testUITests */,
);
};
/* End PBXProject section */
/* Begin PBXResourcesBuildPhase section */
50F787ED2497683900D52237 /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
50F787FF2497683A00D52237 /* LaunchScreen.storyboard in Resources */,
50F787FC2497683A00D52237 /* Preview Assets.xcassets in Resources */,
50F787F92497683A00D52237 /* Assets.xcassets in Resources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
50F788032497683A00D52237 /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
50F7880E2497683A00D52237 /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXResourcesBuildPhase section */
/* Begin PBXSourcesBuildPhase section */
50F787EB2497683900D52237 /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
50F787F32497683900D52237 /* AppDelegate.swift in Sources */,
50F787F52497683900D52237 /* SceneDelegate.swift in Sources */,
50F787F72497683900D52237 /* ContentView.swift in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
50F788012497683A00D52237 /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
50F7880A2497683A00D52237 /* deepspeech_ios_testTests.swift in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
50F7880C2497683A00D52237 /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
50F788152497683A00D52237 /* deepspeech_ios_testUITests.swift in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXSourcesBuildPhase section */
/* Begin PBXTargetDependency section */
50F788072497683A00D52237 /* PBXTargetDependency */ = {
isa = PBXTargetDependency;
target = 50F787EE2497683900D52237 /* deepspeech_ios_test */;
targetProxy = 50F788062497683A00D52237 /* PBXContainerItemProxy */;
};
50F788122497683A00D52237 /* PBXTargetDependency */ = {
isa = PBXTargetDependency;
target = 50F787EE2497683900D52237 /* deepspeech_ios_test */;
targetProxy = 50F788112497683A00D52237 /* PBXContainerItemProxy */;
};
/* End PBXTargetDependency section */
/* Begin PBXVariantGroup section */
50F787FD2497683A00D52237 /* LaunchScreen.storyboard */ = {
isa = PBXVariantGroup;
children = (
50F787FE2497683A00D52237 /* Base */,
);
name = LaunchScreen.storyboard;
sourceTree = "<group>";
};
/* End PBXVariantGroup section */
/* Begin XCBuildConfiguration section */
50F788172497683A00D52237 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++14";
CLANG_CXX_LIBRARY = "libc++";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_ENABLE_OBJC_WEAK = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
COPY_PHASE_STRIP = NO;
DEBUG_INFORMATION_FORMAT = dwarf;
ENABLE_STRICT_OBJC_MSGSEND = YES;
ENABLE_TESTABILITY = YES;
GCC_C_LANGUAGE_STANDARD = gnu11;
GCC_DYNAMIC_NO_PIC = NO;
GCC_NO_COMMON_BLOCKS = YES;
GCC_OPTIMIZATION_LEVEL = 0;
GCC_PREPROCESSOR_DEFINITIONS = (
"DEBUG=1",
"$(inherited)",
);
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
IPHONEOS_DEPLOYMENT_TARGET = 13.5;
MTL_ENABLE_DEBUG_INFO = INCLUDE_SOURCE;
MTL_FAST_MATH = YES;
ONLY_ACTIVE_ARCH = YES;
SDKROOT = iphoneos;
SWIFT_ACTIVE_COMPILATION_CONDITIONS = DEBUG;
SWIFT_OPTIMIZATION_LEVEL = "-Onone";
};
name = Debug;
};
50F788182497683A00D52237 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++14";
CLANG_CXX_LIBRARY = "libc++";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_ENABLE_OBJC_WEAK = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
COPY_PHASE_STRIP = NO;
DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
ENABLE_NS_ASSERTIONS = NO;
ENABLE_STRICT_OBJC_MSGSEND = YES;
GCC_C_LANGUAGE_STANDARD = gnu11;
GCC_NO_COMMON_BLOCKS = YES;
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
IPHONEOS_DEPLOYMENT_TARGET = 13.5;
MTL_ENABLE_DEBUG_INFO = NO;
MTL_FAST_MATH = YES;
SDKROOT = iphoneos;
SWIFT_COMPILATION_MODE = wholemodule;
SWIFT_OPTIMIZATION_LEVEL = "-O";
VALIDATE_PRODUCT = YES;
};
name = Release;
};
50F7881A2497683A00D52237 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
CODE_SIGN_STYLE = Automatic;
DEVELOPMENT_ASSET_PATHS = "\"deepspeech_ios_test/Preview Content\"";
DEVELOPMENT_TEAM = AWCG9S27P7;
ENABLE_PREVIEWS = YES;
FRAMEWORK_SEARCH_PATHS = "$(PROJECT_DIR)";
INFOPLIST_FILE = deepspeech_ios_test/Info.plist;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
);
LIBRARY_SEARCH_PATHS = (
"$(inherited)",
"$(PROJECT_DIR)",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios-test";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Debug;
};
50F7881B2497683A00D52237 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
CODE_SIGN_STYLE = Automatic;
DEVELOPMENT_ASSET_PATHS = "\"deepspeech_ios_test/Preview Content\"";
DEVELOPMENT_TEAM = AWCG9S27P7;
ENABLE_PREVIEWS = YES;
FRAMEWORK_SEARCH_PATHS = "$(PROJECT_DIR)";
INFOPLIST_FILE = deepspeech_ios_test/Info.plist;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
);
LIBRARY_SEARCH_PATHS = (
"$(inherited)",
"$(PROJECT_DIR)",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios-test";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
};
name = Release;
};
50F7881D2497683A00D52237 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_EMBED_SWIFT_STANDARD_LIBRARIES = YES;
BUNDLE_LOADER = "$(TEST_HOST)";
CODE_SIGN_STYLE = Automatic;
INFOPLIST_FILE = deepspeech_ios_testTests/Info.plist;
IPHONEOS_DEPLOYMENT_TARGET = 13.5;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios-testTests";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
TEST_HOST = "$(BUILT_PRODUCTS_DIR)/deepspeech_ios_test.app/deepspeech_ios_test";
};
name = Debug;
};
50F7881E2497683A00D52237 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_EMBED_SWIFT_STANDARD_LIBRARIES = YES;
BUNDLE_LOADER = "$(TEST_HOST)";
CODE_SIGN_STYLE = Automatic;
INFOPLIST_FILE = deepspeech_ios_testTests/Info.plist;
IPHONEOS_DEPLOYMENT_TARGET = 13.5;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios-testTests";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
TEST_HOST = "$(BUILT_PRODUCTS_DIR)/deepspeech_ios_test.app/deepspeech_ios_test";
};
name = Release;
};
50F788202497683A00D52237 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_EMBED_SWIFT_STANDARD_LIBRARIES = YES;
CODE_SIGN_STYLE = Automatic;
INFOPLIST_FILE = deepspeech_ios_testUITests/Info.plist;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios-testUITests";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
TEST_TARGET_NAME = deepspeech_ios_test;
};
name = Debug;
};
50F788212497683A00D52237 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_EMBED_SWIFT_STANDARD_LIBRARIES = YES;
CODE_SIGN_STYLE = Automatic;
INFOPLIST_FILE = deepspeech_ios_testUITests/Info.plist;
LD_RUNPATH_SEARCH_PATHS = (
"$(inherited)",
"@executable_path/Frameworks",
"@loader_path/Frameworks",
);
PRODUCT_BUNDLE_IDENTIFIER = "org.mozilla.deepspeech-ios-testUITests";
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 5.0;
TARGETED_DEVICE_FAMILY = "1,2";
TEST_TARGET_NAME = deepspeech_ios_test;
};
name = Release;
};
/* End XCBuildConfiguration section */
/* Begin XCConfigurationList section */
50F787EA2497683900D52237 /* Build configuration list for PBXProject "deepspeech_ios_test" */ = {
isa = XCConfigurationList;
buildConfigurations = (
50F788172497683A00D52237 /* Debug */,
50F788182497683A00D52237 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
50F788192497683A00D52237 /* Build configuration list for PBXNativeTarget "deepspeech_ios_test" */ = {
isa = XCConfigurationList;
buildConfigurations = (
50F7881A2497683A00D52237 /* Debug */,
50F7881B2497683A00D52237 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
50F7881C2497683A00D52237 /* Build configuration list for PBXNativeTarget "deepspeech_ios_testTests" */ = {
isa = XCConfigurationList;
buildConfigurations = (
50F7881D2497683A00D52237 /* Debug */,
50F7881E2497683A00D52237 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
50F7881F2497683A00D52237 /* Build configuration list for PBXNativeTarget "deepspeech_ios_testUITests" */ = {
isa = XCConfigurationList;
buildConfigurations = (
50F788202497683A00D52237 /* Debug */,
50F788212497683A00D52237 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
/* End XCConfigurationList section */
};
rootObject = 50F787E72497683900D52237 /* Project object */;
}

7
native_client/swift/deepspeech_ios_test.xcodeproj/project.xcworkspace/contents.xcworkspacedata generated Normal file
View File

@ -0,0 +1,7 @@
<?xml version="1.0" encoding="UTF-8"?>
<Workspace
version = "1.0">
<FileRef
location = "self:deepspeech_ios_test.xcodeproj">
</FileRef>
</Workspace>

8
native_client/swift/deepspeech_ios_test.xcodeproj/project.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist Normal file
View File

@ -0,0 +1,8 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>IDEDidComputeMac32BitWarning</key>
<true/>
</dict>
</plist>

106
native_client/swift/deepspeech_ios_test.xcodeproj/xcshareddata/xcschemes/deepspeech_ios_test.xcscheme Normal file
View File

@ -0,0 +1,106 @@
<?xml version="1.0" encoding="UTF-8"?>
<Scheme
LastUpgradeVersion = "1150"
version = "1.3">
<BuildAction
parallelizeBuildables = "YES"
buildImplicitDependencies = "YES">
<BuildActionEntries>
<BuildActionEntry
buildForTesting = "YES"
buildForRunning = "YES"
buildForProfiling = "YES"
buildForArchiving = "YES"
buildForAnalyzing = "YES">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "50F787EE2497683900D52237"
BuildableName = "deepspeech_ios_test.app"
BlueprintName = "deepspeech_ios_test"
ReferencedContainer = "container:deepspeech_ios_test.xcodeproj">
</BuildableReference>
</BuildActionEntry>
</BuildActionEntries>
</BuildAction>
<TestAction
buildConfiguration = "Debug"
selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
shouldUseLaunchSchemeArgsEnv = "YES">
<Testables>
<TestableReference
skipped = "NO">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "50F788042497683A00D52237"
BuildableName = "deepspeech_ios_testTests.xctest"
BlueprintName = "deepspeech_ios_testTests"
ReferencedContainer = "container:deepspeech_ios_test.xcodeproj">
</BuildableReference>
</TestableReference>
<TestableReference
skipped = "NO">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "50F7880F2497683A00D52237"
BuildableName = "deepspeech_ios_testUITests.xctest"
BlueprintName = "deepspeech_ios_testUITests"
ReferencedContainer = "container:deepspeech_ios_test.xcodeproj">
</BuildableReference>
</TestableReference>
</Testables>
</TestAction>
<LaunchAction
buildConfiguration = "Debug"
selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
enableAddressSanitizer = "YES"
launchStyle = "0"
useCustomWorkingDirectory = "NO"
ignoresPersistentStateOnLaunch = "NO"
debugDocumentVersioning = "YES"
debugServiceExtension = "internal"
allowLocationSimulation = "YES">
<BuildableProductRunnable
runnableDebuggingMode = "0">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "50F787EE2497683900D52237"
BuildableName = "deepspeech_ios_test.app"
BlueprintName = "deepspeech_ios_test"
ReferencedContainer = "container:deepspeech_ios_test.xcodeproj">
</BuildableReference>
</BuildableProductRunnable>
<AdditionalOptions>
<AdditionalOption
key = "NSZombieEnabled"
value = "YES"
isEnabled = "YES">
</AdditionalOption>
</AdditionalOptions>
</LaunchAction>
<ProfileAction
buildConfiguration = "Release"
shouldUseLaunchSchemeArgsEnv = "YES"
savedToolIdentifier = ""
useCustomWorkingDirectory = "NO"
debugDocumentVersioning = "YES">
<BuildableProductRunnable
runnableDebuggingMode = "0">
<BuildableReference
BuildableIdentifier = "primary"
BlueprintIdentifier = "50F787EE2497683900D52237"
BuildableName = "deepspeech_ios_test.app"
BlueprintName = "deepspeech_ios_test"
ReferencedContainer = "container:deepspeech_ios_test.xcodeproj">
</BuildableReference>
</BuildableProductRunnable>
</ProfileAction>
<AnalyzeAction
buildConfiguration = "Debug">
</AnalyzeAction>
<ArchiveAction
buildConfiguration = "Release"
revealArchiveInOrganizer = "YES">
</ArchiveAction>
</Scheme>

210
native_client/swift/deepspeech_ios_test/AppDelegate.swift Normal file
View File

@ -0,0 +1,210 @@
//
// AppDelegate.swift
// deepspeech_ios_test
//
// Created by Reuben Morais on 15.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
import UIKit
import Foundation
import AVFoundation
import AudioToolbox
import Accelerate
import deepspeech_ios
/// Holds audio information used for building waveforms
final class AudioContext {
/// The audio asset URL used to load the context
public let audioURL: URL
/// Total number of samples in loaded asset
public let totalSamples: Int
/// Loaded asset
public let asset: AVAsset
// Loaded assetTrack
public let assetTrack: AVAssetTrack
private init(audioURL: URL, totalSamples: Int, asset: AVAsset, assetTrack: AVAssetTrack) {
self.audioURL = audioURL
self.totalSamples = totalSamples
self.asset = asset
self.assetTrack = assetTrack
}
public static func load(fromAudioURL audioURL: URL, completionHandler: @escaping (_ audioContext: AudioContext?) -> ()) {
let asset = AVURLAsset(url: audioURL, options: [AVURLAssetPreferPreciseDurationAndTimingKey: NSNumber(value: true as Bool)])
guard let assetTrack = asset.tracks(withMediaType: AVMediaType.audio).first else {
fatalError("Couldn't load AVAssetTrack")
}
asset.loadValuesAsynchronously(forKeys: ["duration"]) {
var error: NSError?
let status = asset.statusOfValue(forKey: "duration", error: &error)
switch status {
case .loaded:
guard
let formatDescriptions = assetTrack.formatDescriptions as? [CMAudioFormatDescription],
let audioFormatDesc = formatDescriptions.first,
let asbd = CMAudioFormatDescriptionGetStreamBasicDescription(audioFormatDesc)
else { break }
let totalSamples = Int((asbd.pointee.mSampleRate) * Float64(asset.duration.value) / Float64(asset.duration.timescale))
let audioContext = AudioContext(audioURL: audioURL, totalSamples: totalSamples, asset: asset, assetTrack: assetTrack)
completionHandler(audioContext)
return
case .failed, .cancelled, .loading, .unknown:
print("Couldn't load asset: \(error?.localizedDescription ?? "Unknown error")")
}
completionHandler(nil)
}
}
}
func render(audioContext: AudioContext?, stream: DeepSpeechStream) {
guard let audioContext = audioContext else {
fatalError("Couldn't create the audioContext")
}
let sampleRange: CountableRange<Int> = 0..<audioContext.totalSamples
guard let reader = try? AVAssetReader(asset: audioContext.asset)
else {
fatalError("Couldn't initialize the AVAssetReader")
}
reader.timeRange = CMTimeRange(start: CMTime(value: Int64(sampleRange.lowerBound), timescale: audioContext.asset.duration.timescale),
duration: CMTime(value: Int64(sampleRange.count), timescale: audioContext.asset.duration.timescale))
let outputSettingsDict: [String : Any] = [
AVFormatIDKey: Int(kAudioFormatLinearPCM),
AVLinearPCMBitDepthKey: 16,
AVLinearPCMIsBigEndianKey: false,
AVLinearPCMIsFloatKey: false,
AVLinearPCMIsNonInterleaved: false
]
let readerOutput = AVAssetReaderTrackOutput(track: audioContext.assetTrack,
outputSettings: outputSettingsDict)
readerOutput.alwaysCopiesSampleData = false
reader.add(readerOutput)
var sampleBuffer = Data()
// 16-bit samples
reader.startReading()
defer { reader.cancelReading() }
while reader.status == .reading {
guard let readSampleBuffer = readerOutput.copyNextSampleBuffer(),
let readBuffer = CMSampleBufferGetDataBuffer(readSampleBuffer) else {
break
}
// Append audio sample buffer into our current sample buffer
var readBufferLength = 0
var readBufferPointer: UnsafeMutablePointer<Int8>?
CMBlockBufferGetDataPointer(readBuffer,
atOffset: 0,
lengthAtOffsetOut: &readBufferLength,
totalLengthOut: nil,
dataPointerOut: &readBufferPointer)
sampleBuffer.append(UnsafeBufferPointer(start: readBufferPointer, count: readBufferLength))
CMSampleBufferInvalidate(readSampleBuffer)
let totalSamples = sampleBuffer.count / MemoryLayout<Int16>.size
print("read \(totalSamples) samples")
sampleBuffer.withUnsafeBytes { (samples: UnsafeRawBufferPointer) in
let unsafeBufferPointer = samples.bindMemory(to: Int16.self)
stream.feedAudioContent(buffer: unsafeBufferPointer)
}
sampleBuffer.removeAll()
}
// if (reader.status == AVAssetReaderStatusFailed || reader.status == AVAssetReaderStatusUnknown)
guard reader.status == .completed else {
fatalError("Couldn't read the audio file")
}
}
func test(model: DeepSpeechModel, audioPath: String, completion: @escaping () -> ()) {
let url = URL(fileURLWithPath: audioPath)
let stream = try! model.createStream()
print("\(audioPath)")
let start = CFAbsoluteTimeGetCurrent()
AudioContext.load(fromAudioURL: url, completionHandler: { audioContext in
guard let audioContext = audioContext else {
fatalError("Couldn't create the audioContext")
}
render(audioContext: audioContext, stream: stream)
let result = stream.finishStream()
let end = CFAbsoluteTimeGetCurrent()
print("\"\(audioPath)\": \(end - start) - \(result)")
completion()
})
}
@UIApplicationMain
class AppDelegate: UIResponder, UIApplicationDelegate {
func application(_ application: UIApplication, didFinishLaunchingWithOptions launchOptions: [UIApplication.LaunchOptionsKey: Any]?) -> Bool {
let model = try! DeepSpeechModel(modelPath: Bundle.main.path(forResource: "output_graph", ofType: "tflite")!)
try! model.enableExternalScorer(scorerPath: Bundle.main.path(forResource: "librispeech_en_utf8_nonpruned_o6", ofType: "scorer")!)
let files = [
"5639-40744-0008",
"1089-134686-0019",
"2094-142345-0053",
"8463-294825-0010",
"121-123852-0001",
"7021-79740-0008",
"6930-76324-0010",
"5105-28240-0001",
"1089-134691-0012",
"5142-33396-0027",
"260-123288-0004",
"6930-75918-0008",
"8463-294828-0005",
"61-70970-0002"
]
let serialQueue = DispatchQueue(label: "serialQueue")
let group = DispatchGroup()
group.enter()
serialQueue.async {
test(model: model, audioPath: Bundle.main.path(forResource: "1284-134647-0003", ofType: "wav")!) {
group.leave()
}
}
for path in files {
group.wait()
group.enter()
test(model: model, audioPath: Bundle.main.path(forResource: path, ofType: "wav")!) {
group.leave()
}
}
return true
}
// MARK: UISceneSession Lifecycle
func application(_ application: UIApplication, configurationForConnecting connectingSceneSession: UISceneSession, options: UIScene.ConnectionOptions) -> UISceneConfiguration {
// Called when a new scene session is being created.
// Use this method to select a configuration to create the new scene with.
return UISceneConfiguration(name: "Default Configuration", sessionRole: connectingSceneSession.role)
}
func application(_ application: UIApplication, didDiscardSceneSessions sceneSessions: Set<UISceneSession>) {
// Called when the user discards a scene session.
// If any sessions were discarded while the application was not running, this will be called shortly after application:didFinishLaunchingWithOptions.
// Use this method to release any resources that were specific to the discarded scenes, as they will not return.
}
}

98
native_client/swift/deepspeech_ios_test/Assets.xcassets/AppIcon.appiconset/Contents.json Normal file
View File

@ -0,0 +1,98 @@
{
"images" : [
{
"idiom" : "iphone",
"scale" : "2x",
"size" : "20x20"
},
{
"idiom" : "iphone",
"scale" : "3x",
"size" : "20x20"
},
{
"idiom" : "iphone",
"scale" : "2x",
"size" : "29x29"
},
{
"idiom" : "iphone",
"scale" : "3x",
"size" : "29x29"
},
{
"idiom" : "iphone",
"scale" : "2x",
"size" : "40x40"
},
{
"idiom" : "iphone",
"scale" : "3x",
"size" : "40x40"
},
{
"idiom" : "iphone",
"scale" : "2x",
"size" : "60x60"
},
{
"idiom" : "iphone",
"scale" : "3x",
"size" : "60x60"
},
{
"idiom" : "ipad",
"scale" : "1x",
"size" : "20x20"
},
{
"idiom" : "ipad",
"scale" : "2x",
"size" : "20x20"
},
{
"idiom" : "ipad",
"scale" : "1x",
"size" : "29x29"
},
{
"idiom" : "ipad",
"scale" : "2x",
"size" : "29x29"
},
{
"idiom" : "ipad",
"scale" : "1x",
"size" : "40x40"
},
{
"idiom" : "ipad",
"scale" : "2x",
"size" : "40x40"
},
{
"idiom" : "ipad",
"scale" : "1x",
"size" : "76x76"
},
{
"idiom" : "ipad",
"scale" : "2x",
"size" : "76x76"
},
{
"idiom" : "ipad",
"scale" : "2x",
"size" : "83.5x83.5"
},
{
"idiom" : "ios-marketing",
"scale" : "1x",
"size" : "1024x1024"
}
],
"info" : {
"author" : "xcode",
"version" : 1
}
}

6
native_client/swift/deepspeech_ios_test/Assets.xcassets/Contents.json Normal file
View File

@ -0,0 +1,6 @@
{
"info" : {
"author" : "xcode",
"version" : 1
}
}

25
native_client/swift/deepspeech_ios_test/Base.lproj/LaunchScreen.storyboard Normal file
View File

@ -0,0 +1,25 @@
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<document type="com.apple.InterfaceBuilder3.CocoaTouch.Storyboard.XIB" version="3.0" toolsVersion="13122.16" targetRuntime="iOS.CocoaTouch" propertyAccessControl="none" useAutolayout="YES" launchScreen="YES" useTraitCollections="YES" useSafeAreas="YES" colorMatched="YES" initialViewController="01J-lp-oVM">
<dependencies>
<plugIn identifier="com.apple.InterfaceBuilder.IBCocoaTouchPlugin" version="13104.12"/>
<capability name="Safe area layout guides" minToolsVersion="9.0"/>
<capability name="documents saved in the Xcode 8 format" minToolsVersion="8.0"/>
</dependencies>
<scenes>
<!--View Controller-->
<scene sceneID="EHf-IW-A2E">
<objects>
<viewController id="01J-lp-oVM" sceneMemberID="viewController">
<view key="view" contentMode="scaleToFill" id="Ze5-6b-2t3">
<rect key="frame" x="0.0" y="0.0" width="375" height="667"/>
<autoresizingMask key="autoresizingMask" widthSizable="YES" heightSizable="YES"/>
<color key="backgroundColor" xcode11CocoaTouchSystemColor="systemBackgroundColor" cocoaTouchSystemColor="whiteColor"/>
<viewLayoutGuide key="safeArea" id="6Tk-OE-BBY"/>
</view>
</viewController>
<placeholder placeholderIdentifier="IBFirstResponder" id="iYj-Kq-Ea1" userLabel="First Responder" sceneMemberID="firstResponder"/>
</objects>
<point key="canvasLocation" x="53" y="375"/>
</scene>
</scenes>
</document>

21
native_client/swift/deepspeech_ios_test/ContentView.swift Normal file
View File

@ -0,0 +1,21 @@
//
// ContentView.swift
// deepspeech_ios_test
//
// Created by Reuben Morais on 15.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
import SwiftUI
struct ContentView: View {
var body: some View {
Text("Hello, World!")
}
}
struct ContentView_Previews: PreviewProvider {
static var previews: some View {
ContentView()
}
}

60
native_client/swift/deepspeech_ios_test/Info.plist Normal file
View File

@ -0,0 +1,60 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>$(DEVELOPMENT_LANGUAGE)</string>
<key>CFBundleExecutable</key>
<string>$(EXECUTABLE_NAME)</string>
<key>CFBundleIdentifier</key>
<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>$(PRODUCT_NAME)</string>
<key>CFBundlePackageType</key>
<string>$(PRODUCT_BUNDLE_PACKAGE_TYPE)</string>
<key>CFBundleShortVersionString</key>
<string>1.0</string>
<key>CFBundleVersion</key>
<string>1</string>
<key>LSRequiresIPhoneOS</key>
<true/>
<key>UIApplicationSceneManifest</key>
<dict>
<key>UIApplicationSupportsMultipleScenes</key>
<false/>
<key>UISceneConfigurations</key>
<dict>
<key>UIWindowSceneSessionRoleApplication</key>
<array>
<dict>
<key>UISceneConfigurationName</key>
<string>Default Configuration</string>
<key>UISceneDelegateClassName</key>
<string>$(PRODUCT_MODULE_NAME).SceneDelegate</string>
</dict>
</array>
</dict>
</dict>
<key>UILaunchStoryboardName</key>
<string>LaunchScreen</string>
<key>UIRequiredDeviceCapabilities</key>
<array>
<string>armv7</string>
</array>
<key>UISupportedInterfaceOrientations</key>
<array>
<string>UIInterfaceOrientationPortrait</string>
<string>UIInterfaceOrientationLandscapeLeft</string>
<string>UIInterfaceOrientationLandscapeRight</string>
</array>
<key>UISupportedInterfaceOrientations~ipad</key>
<array>
<string>UIInterfaceOrientationPortrait</string>
<string>UIInterfaceOrientationPortraitUpsideDown</string>
<string>UIInterfaceOrientationLandscapeLeft</string>
<string>UIInterfaceOrientationLandscapeRight</string>
</array>
</dict>
</plist>

6
native_client/swift/deepspeech_ios_test/Preview Content/Preview Assets.xcassets/Contents.json Normal file
View File

@ -0,0 +1,6 @@
{
"info" : {
"author" : "xcode",
"version" : 1
}
}

64
native_client/swift/deepspeech_ios_test/SceneDelegate.swift Normal file
View File

@ -0,0 +1,64 @@
//
// SceneDelegate.swift
// deepspeech_ios_test
//
// Created by Reuben Morais on 15.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
import UIKit
import SwiftUI
class SceneDelegate: UIResponder, UIWindowSceneDelegate {
var window: UIWindow?
func scene(_ scene: UIScene, willConnectTo session: UISceneSession, options connectionOptions: UIScene.ConnectionOptions) {
// Use this method to optionally configure and attach the UIWindow `window` to the provided UIWindowScene `scene`.
// If using a storyboard, the `window` property will automatically be initialized and attached to the scene.
// This delegate does not imply the connecting scene or session are new (see `application:configurationForConnectingSceneSession` instead).
// Create the SwiftUI view that provides the window contents.
let contentView = ContentView()
// Use a UIHostingController as window root view controller.
if let windowScene = scene as? UIWindowScene {
let window = UIWindow(windowScene: windowScene)
window.rootViewController = UIHostingController(rootView: contentView)
self.window = window
window.makeKeyAndVisible()
}
}
func sceneDidDisconnect(_ scene: UIScene) {
// Called as the scene is being released by the system.
// This occurs shortly after the scene enters the background, or when its session is discarded.
// Release any resources associated with this scene that can be re-created the next time the scene connects.
// The scene may re-connect later, as its session was not neccessarily discarded (see `application:didDiscardSceneSessions` instead).
}
func sceneDidBecomeActive(_ scene: UIScene) {
// Called when the scene has moved from an inactive state to an active state.
// Use this method to restart any tasks that were paused (or not yet started) when the scene was inactive.
}
func sceneWillResignActive(_ scene: UIScene) {
// Called when the scene will move from an active state to an inactive state.
// This may occur due to temporary interruptions (ex. an incoming phone call).
}
func sceneWillEnterForeground(_ scene: UIScene) {
// Called as the scene transitions from the background to the foreground.
// Use this method to undo the changes made on entering the background.
}
func sceneDidEnterBackground(_ scene: UIScene) {
// Called as the scene transitions from the foreground to the background.
// Use this method to save data, release shared resources, and store enough scene-specific state information
// to restore the scene back to its current state.
}
}

22
native_client/swift/deepspeech_ios_testTests/Info.plist Normal file
View File

@ -0,0 +1,22 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>$(DEVELOPMENT_LANGUAGE)</string>
<key>CFBundleExecutable</key>
<string>$(EXECUTABLE_NAME)</string>
<key>CFBundleIdentifier</key>
<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>$(PRODUCT_NAME)</string>
<key>CFBundlePackageType</key>
<string>$(PRODUCT_BUNDLE_PACKAGE_TYPE)</string>
<key>CFBundleShortVersionString</key>
<string>1.0</string>
<key>CFBundleVersion</key>
<string>1</string>
</dict>
</plist>

34
native_client/swift/deepspeech_ios_testTests/deepspeech_ios_testTests.swift Normal file
View File

@ -0,0 +1,34 @@
//
// deepspeech_ios_testTests.swift
// deepspeech_ios_testTests
//
// Created by Reuben Morais on 15.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
import XCTest
@testable import deepspeech_ios_test
class deepspeech_ios_testTests: XCTestCase {
override func setUpWithError() throws {
// Put setup code here. This method is called before the invocation of each test method in the class.
}
override func tearDownWithError() throws {
// Put teardown code here. This method is called after the invocation of each test method in the class.
}
func testExample() throws {
// This is an example of a functional test case.
// Use XCTAssert and related functions to verify your tests produce the correct results.
}
func testPerformanceExample() throws {
// This is an example of a performance test case.
self.measure {
// Put the code you want to measure the time of here.
}
}
}

22
native_client/swift/deepspeech_ios_testUITests/Info.plist Normal file
View File

@ -0,0 +1,22 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>$(DEVELOPMENT_LANGUAGE)</string>
<key>CFBundleExecutable</key>
<string>$(EXECUTABLE_NAME)</string>
<key>CFBundleIdentifier</key>
<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>$(PRODUCT_NAME)</string>
<key>CFBundlePackageType</key>
<string>$(PRODUCT_BUNDLE_PACKAGE_TYPE)</string>
<key>CFBundleShortVersionString</key>
<string>1.0</string>
<key>CFBundleVersion</key>
<string>1</string>
</dict>
</plist>

43
native_client/swift/deepspeech_ios_testUITests/deepspeech_ios_testUITests.swift Normal file
View File

@ -0,0 +1,43 @@
//
// deepspeech_ios_testUITests.swift
// deepspeech_ios_testUITests
//
// Created by Reuben Morais on 15.06.20.
// Copyright © 2020 Mozilla. All rights reserved.
//
import XCTest
class deepspeech_ios_testUITests: XCTestCase {
override func setUpWithError() throws {
// Put setup code here. This method is called before the invocation of each test method in the class.
// In UI tests it is usually best to stop immediately when a failure occurs.
continueAfterFailure = false
// In UI tests it’s important to set the initial state - such as interface orientation - required for your tests before they run. The setUp method is a good place to do this.
}
override func tearDownWithError() throws {
// Put teardown code here. This method is called after the invocation of each test method in the class.
}
func testExample() throws {
// UI tests must launch the application that they test.
let app = XCUIApplication()
app.launch()
// Use recording to get started writing UI tests.
// Use XCTAssert and related functions to verify your tests produce the correct results.
}
func testLaunchPerformance() throws {
if #available(macOS 10.15, iOS 13.0, tvOS 13.0, *) {
// This measures how long it takes to launch your application.
measure(metrics: [XCTOSSignpostMetric.applicationLaunch]) {
XCUIApplication().launch()
}
}
}
}

2
taskcluster/.build.yml
View File

@ -25,7 +25,7 @@ build:
nc_asset_name: 'native_client.tar.xz' nc_asset_name: 'native_client.tar.xz'
args: args:
tests_cmdline: '' tests_cmdline: ''
tensorflow_git_desc: 'TensorFlow: v2.2.0-15-g518c1d0' tensorflow_git_desc: 'TensorFlow: v2.2.0-17-g0854bb5'
test_model_task: '' test_model_task: ''
homebrew: homebrew:
url: '' url: ''

56
taskcluster/.shared.yml
View File

@ -14,8 +14,6 @@ deepspeech:
tensorflow: tensorflow:
packages_xenial: packages_xenial:
apt: 'apt-get -qq update && apt-get -qq -y install realpath build-essential python-virtualenv python-dev python-pip libblas-dev liblapack-dev gfortran wget software-properties-common pixz zip zlib1g-dev unzip' apt: 'apt-get -qq update && apt-get -qq -y install realpath build-essential python-virtualenv python-dev python-pip libblas-dev liblapack-dev gfortran wget software-properties-common pixz zip zlib1g-dev unzip'
packages_macos:
brew: '$TASKCLUSTER_TASK_DIR/DeepSpeech/ds/taskcluster/tf_tc-brew.sh'
packages_win: packages_win:
pacman: 'pacman --noconfirm -S patch unzip tar' pacman: 'pacman --noconfirm -S patch unzip tar'
msys64: 'ln -s $USERPROFILE/msys64 $TASKCLUSTER_TASK_DIR/msys64' msys64: 'ln -s $USERPROFILE/msys64 $TASKCLUSTER_TASK_DIR/msys64'
@ -68,11 +66,11 @@ system:
url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.node-gyp-cache.7/artifacts/public/node-gyp-cache.tar.gz' url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.node-gyp-cache.7/artifacts/public/node-gyp-cache.tar.gz'
namespace: 'project.deepspeech.node-gyp-cache.7' namespace: 'project.deepspeech.node-gyp-cache.7'
homebrew_builds: homebrew_builds:
url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.homebrew_builds.7/artifacts/public/homebrew_builds.tar.gz' url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.homebrew_builds.8/artifacts/public/homebrew_builds.tar.gz'
namespace: 'project.deepspeech.homebrew_builds.7' namespace: 'project.deepspeech.homebrew_builds.8'
homebrew_tests: homebrew_tests:
url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.homebrew_tests.8/artifacts/public/homebrew_tests.tar.gz' url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.homebrew_tests.9/artifacts/public/homebrew_tests.tar.gz'
namespace: 'project.deepspeech.homebrew_tests.8' namespace: 'project.deepspeech.homebrew_tests.9'
android_cache: android_cache:
arm64_v8a: arm64_v8a:
android_24: android_24:
@ -122,8 +120,8 @@ system:
url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.pyenv.linux.8/artifacts/public/pyenv.tar.gz' url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.pyenv.linux.8/artifacts/public/pyenv.tar.gz'
namespace: 'project.deepspeech.pyenv.linux.8' namespace: 'project.deepspeech.pyenv.linux.8'
osx: osx:
url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.pyenv.osx.8/artifacts/public/pyenv.tar.gz' url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.pyenv.osx.9/artifacts/public/pyenv.tar.gz'
namespace: 'project.deepspeech.pyenv.osx.8' namespace: 'project.deepspeech.pyenv.osx.9'
win: win:
url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.pyenv.win.8/artifacts/public/pyenv.tar.gz' url: 'https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.pyenv.win.8/artifacts/public/pyenv.tar.gz'
namespace: 'project.deepspeech.pyenv.win.8' namespace: 'project.deepspeech.pyenv.win.8'
@ -142,32 +140,38 @@ system:
namespace: "project.deepspeech.swig.win.amd64.b5fea54d39832d1d132d7dd921b69c0c2c9d5118" namespace: "project.deepspeech.swig.win.amd64.b5fea54d39832d1d132d7dd921b69c0c2c9d5118"
tensorflow: tensorflow:
linux_amd64_cpu: linux_amd64_cpu:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.1.cpu/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.cpu/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.1.cpu" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.cpu"
linux_amd64_cuda: linux_amd64_cuda:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.cuda/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.cuda/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.cuda" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.cuda"
linux_armv7: linux_armv7:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.arm/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.arm/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.arm" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.arm"
linux_arm64: linux_arm64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.arm64/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.arm64/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.arm64" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.arm64"
darwin_amd64: darwin_amd64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.osx/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.3.osx/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.0.osx" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.3.osx"
android_arm64: android_arm64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1.1.bf55d362bb11e973b8f5.1.a3e5bf44d.1.android-arm64/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.android-arm64/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1.1.bf55d362bb11e973b8f5.1.a3e5bf44d.1.android-arm64" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.android-arm64"
android_armv7: android_armv7:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1.1.bf55d362bb11e973b8f5.1.a3e5bf44d.1.android-armv7/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.android-armv7/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1.1.bf55d362bb11e973b8f5.1.a3e5bf44d.1.android-armv7" namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.android-armv7"
win_amd64_cpu: win_amd64_cpu:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.1.win/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.2.win/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.1.win" namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.2.win"
win_amd64_cuda: win_amd64_cuda:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.1.win-cuda/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.2.win-cuda/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.1.win-cuda" namespace: "project.deepspeech.tensorflow.pip.r2.2.518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d.2.win-cuda"
ios_arm64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.3.ios_arm64/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.3.ios_arm64"
ios_x86_64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.3.ios_x86_64/artifacts/public/home.tar.xz"
namespace: "project.deepspeech.tensorflow.pip.r2.2.0854bb5188a3150a4d75a1c71ee610b0d45cfcb1.3.ios_x86_64"
username: 'build-user' username: 'build-user'
homedir: homedir:
linux: '/home/build-user' linux: '/home/build-user'

7
taskcluster/darwin-opt-base.tyml
View File

@ -35,11 +35,8 @@ payload:
# There is no VM yet running tasks on OSX # There is no VM yet running tasks on OSX
# so one should install by hand: # so one should install by hand:
# - brew # - brew
# - xcode (brew would install) # - Xcode 10.1 in /Applications/Xcode.app, then sudo chown -R root:wheel /Applications/Xcode.app
# - brew install gnu-tar # - brew install gnu-tar git pixz wget coreutils pyenv-virtualenv
# - brew install git
# - brew install pixz
# - brew cask install java
# - sudo easy_install pip # - sudo easy_install pip
command: command:

9
taskcluster/homebrew-build.sh
View File

@ -42,12 +42,13 @@ do_prepare_homebrew()
check_homebrew "${_brew_instance}" check_homebrew "${_brew_instance}"
# Force an upgrade to fetch formulae
brew search openssl
# Then we force onto a specific well-known commit # Then we force onto a specific well-known commit
mkdir -p "$(brew --prefix)/Library/Taps/homebrew/homebrew-core"
pushd "$(brew --prefix)/Library/Taps/homebrew/homebrew-core" pushd "$(brew --prefix)/Library/Taps/homebrew/homebrew-core"
git fetch origin && git checkout ${BREW_FORMULAS_COMMIT} git init
git remote add origin https://github.com/Homebrew/homebrew-core.git
git fetch origin
git checkout ${BREW_FORMULAS_COMMIT}
popd popd
} }

2
taskcluster/homebrew_builds-darwin-amd64.yml
View File

@ -9,4 +9,4 @@ build:
package: "taskcluster/homebrew-package.sh --builds" package: "taskcluster/homebrew-package.sh --builds"
metadata: metadata:
name: "Builds Homebrew macOS AMD64" name: "Builds Homebrew macOS AMD64"
description: "Setup a buildsl Homebrew for macOS/AMD64" description: "Setup a builds Homebrew for macOS/AMD64"

21
taskcluster/ios-arm64-tflite-opt.yml Normal file
View File

@ -0,0 +1,21 @@
build:
template_file: darwin-opt-base.tyml
dependencies:
- "swig-darwin-amd64"
- "node-gyp-cache"
- "homebrew_builds-darwin-amd64"
- "pyenv-darwin-amd64"
- "tf_ios-arm64-opt"
routes:
- "index.project.deepspeech.deepspeech.native_client.${event.head.branchortag}.ios_arm64-tflite"
- "index.project.deepspeech.deepspeech.native_client.${event.head.branchortag}.${event.head.sha}.ios_arm64-tflite"
- "index.project.deepspeech.deepspeech.native_client.ios_arm64-tflite.${event.head.sha}"
tensorflow: ${system.tensorflow.ios_arm64.url}
scripts:
build: "taskcluster/ios-build.sh --arm64"
package: "taskcluster/ios-package.sh --arm64"
nc_asset_name: "native_client.arm64.tflite.ios.tar.xz"
maxRunTime: 14400
metadata:
name: "DeepSpeech iOS ARM64 TFLite"
description: "Building DeepSpeech for iOS ARM64, TFLite, optimized version"

25
taskcluster/ios-build.sh Executable file
View File

@ -0,0 +1,25 @@
#!/bin/bash
set -xe
arch=$1
source $(dirname "$0")/tc-tests-utils.sh
source $(dirname "$0")/tf_tc-vars.sh
BAZEL_TARGETS="
//native_client:libdeepspeech.so
"
if [ "${arch}" = "--arm64" ]; then
BAZEL_BUILD_FLAGS="${BAZEL_IOS_ARM64_FLAGS}"
else
BAZEL_BUILD_FLAGS="${BAZEL_IOS_X86_64_FLAGS}"
fi
BAZEL_ENV_FLAGS="TF_NEED_CUDA=0"
do_bazel_build
do_deepspeech_ios_framework_build "${arch}"

30
taskcluster/ios-package.sh Executable file
View File

@ -0,0 +1,30 @@
#!/bin/bash
set -xe
arch=$1
source $(dirname "$0")/tc-tests-utils.sh
mkdir -p ${TASKCLUSTER_ARTIFACTS} || true
cp ${DS_ROOT_TASK}/DeepSpeech/ds/tensorflow/bazel*.log ${TASKCLUSTER_ARTIFACTS}/
package_native_client "native_client.tar.xz"
package_libdeepspeech_as_zip "libdeepspeech.zip"
case $arch in
"--x86_64")
release_folder="Release-iphonesimulator"
artifact_name="deepspeech_ios.framework.x86_64.tar.xz"
;;
"--arm64")
release_folder="Release-iphoneos"
artifact_name="deepspeech_ios.framework.arm64.tar.xz"
;;
esac
${TAR} -cf - \
-C ${DS_ROOT_TASK}/DeepSpeech/ds/native_client/swift/DerivedData/Build/Products/${release_folder}/ deepspeech_ios.framework \
| ${XZ} > "${TASKCLUSTER_ARTIFACTS}/${artifact_name}"

21
taskcluster/ios-x86_64-tflite-opt.yml Normal file
View File

@ -0,0 +1,21 @@
build:
template_file: darwin-opt-base.tyml
dependencies:
- "swig-darwin-amd64"
- "node-gyp-cache"
- "homebrew_builds-darwin-amd64"
- "pyenv-darwin-amd64"
- "tf_ios-x86_64-opt"
routes:
- "index.project.deepspeech.deepspeech.native_client.${event.head.branchortag}.ios_x86_64-tflite"
- "index.project.deepspeech.deepspeech.native_client.${event.head.branchortag}.${event.head.sha}.ios_x86_64-tflite"
- "index.project.deepspeech.deepspeech.native_client.ios_x86_64-tflite.${event.head.sha}"
tensorflow: ${system.tensorflow.ios_x86_64.url}
scripts:
build: "taskcluster/ios-build.sh --x86_64"
package: "taskcluster/ios-package.sh --x86_64"
nc_asset_name: "native_client.x86_64.tflite.ios.tar.xz"
maxRunTime: 14400
metadata:
name: "DeepSpeech iOS x86_64 TFLite"
description: "Building DeepSpeech for iOS x86_64, TFLite, optimized version"

5
taskcluster/scriptworker-task-github.yml
View File

@ -20,6 +20,8 @@ build:
- "win-amd64-cpu-opt" - "win-amd64-cpu-opt"
- "win-amd64-gpu-opt" - "win-amd64-gpu-opt"
- "win-amd64-ctc-opt" - "win-amd64-ctc-opt"
- "ios-x86_64-tflite-opt"
- "ios-arm64-tflite-opt"
allowed: allowed:
- "tag" - "tag"
ref_match: "refs/tags/" ref_match: "refs/tags/"
@ -66,6 +68,9 @@ build:
- "win-amd64-cpu-opt" - "win-amd64-cpu-opt"
- "win-amd64-gpu-opt" - "win-amd64-gpu-opt"
- "win-amd64-tflite-opt" - "win-amd64-tflite-opt"
ios:
- "ios-x86_64-tflite-opt"
- "ios-arm64-tflite-opt"
metadata: metadata:
name: "DeepSpeech GitHub Packages" name: "DeepSpeech GitHub Packages"
description: "Trigger Uploading of DeepSpeech Packages to GitHub release page" description: "Trigger Uploading of DeepSpeech Packages to GitHub release page"

38
taskcluster/tc-build-utils.sh
View File

@ -33,6 +33,8 @@ do_deepspeech_python_build()
virtualenv_activate "${pyalias}" "deepspeech" virtualenv_activate "${pyalias}" "deepspeech"
python --version python --version
pip --version
pip3 --version
which pip which pip
which pip3 which pip3
@ -229,7 +231,7 @@ do_deepspeech_netframework_build()
cd ${DS_DSDIR}/native_client/dotnet cd ${DS_DSDIR}/native_client/dotnet
# Setup dependencies # Setup dependencies
nuget install DeepSpeechConsole/packages.config -OutputDirectory packages/ nuget restore DeepSpeech.sln
MSBUILD="$(cygpath 'C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\MSBuild\Current\Bin\MSBuild.exe')" MSBUILD="$(cygpath 'C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\MSBuild\Current\Bin\MSBuild.exe')"
@ -240,23 +242,30 @@ do_deepspeech_netframework_build()
DeepSpeechClient/DeepSpeechClient.csproj \ DeepSpeechClient/DeepSpeechClient.csproj \
/p:Configuration=Release \ /p:Configuration=Release \
/p:Platform=x64 \ /p:Platform=x64 \
/p:TargetFrameworkVersion="v4.5.2" \ /p:TargetFramework="net452" \
/p:OutputPath=bin/nuget/x64/v4.5 /p:OutputPath=bin/nuget/x64/v4.5
MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \ MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \
DeepSpeechClient/DeepSpeechClient.csproj \ DeepSpeechClient/DeepSpeechClient.csproj \
/p:Configuration=Release \ /p:Configuration=Release \
/p:Platform=x64 \ /p:Platform=x64 \
/p:TargetFrameworkVersion="v4.6" \ /p:TargetFramework="net46" \
/p:OutputPath=bin/nuget/x64/v4.6 /p:OutputPath=bin/nuget/x64/v4.6
MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \ MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \
DeepSpeechClient/DeepSpeechClient.csproj \ DeepSpeechClient/DeepSpeechClient.csproj \
/p:Configuration=Release \ /p:Configuration=Release \
/p:Platform=x64 \ /p:Platform=x64 \
/p:TargetFrameworkVersion="v4.7" \ /p:TargetFramework="net47" \
/p:OutputPath=bin/nuget/x64/v4.7 /p:OutputPath=bin/nuget/x64/v4.7
MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \
DeepSpeechClient/DeepSpeechClient.csproj \
/p:Configuration=Release \
/p:Platform=x64 \
/p:TargetFramework="uap10.0" \
/p:OutputPath=bin/nuget/x64/uap10.0
MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \ MSYS2_ARG_CONV_EXCL='/' "${MSBUILD}" \
DeepSpeechConsole/DeepSpeechConsole.csproj \ DeepSpeechConsole/DeepSpeechConsole.csproj \
/p:Configuration=Release \ /p:Configuration=Release \
@ -305,6 +314,9 @@ do_nuget_build()
mkdir -p nupkg/lib/net47/ mkdir -p nupkg/lib/net47/
cp DeepSpeechClient/bin/nuget/x64/v4.7/DeepSpeechClient.dll nupkg/lib/net47/ cp DeepSpeechClient/bin/nuget/x64/v4.7/DeepSpeechClient.dll nupkg/lib/net47/
mkdir -p nupkg/lib/uap10.0/
cp DeepSpeechClient/bin/nuget/x64/uap10.0/DeepSpeechClient.dll nupkg/lib/uap10.0/
PROJECT_VERSION=$(strip "${DS_VERSION}") PROJECT_VERSION=$(strip "${DS_VERSION}")
sed \ sed \
-e "s/\$NUPKG_ID/${PROJECT_NAME}/" \ -e "s/\$NUPKG_ID/${PROJECT_NAME}/" \
@ -313,3 +325,21 @@ do_nuget_build()
nuget pack nupkg/deepspeech.nuspec nuget pack nupkg/deepspeech.nuspec
} }
do_deepspeech_ios_framework_build()
{
arch=$1
cp ${DS_TFDIR}/bazel-bin/native_client/libdeepspeech.so ${DS_DSDIR}/native_client/swift/libdeepspeech.so
cd ${DS_DSDIR}/native_client/swift
case $arch in
"--x86_64")
iosSDK="iphonesimulator"
xcodeArch="x86_64"
;;
"--arm64")
iosSDK="iphoneos"
xcodeArch="arm64"
;;
esac
xcodebuild -workspace deepspeech_ios.xcworkspace -scheme deepspeech_ios_test -configuration Release -arch "${xcodeArch}" -sdk "${iosSDK}" -derivedDataPath DerivedData CODE_SIGN_IDENTITY="" CODE_SIGNING_REQUIRED=NO CODE_SIGNING_ALLOWED=NO
}

4
taskcluster/tc-cpp-ds-tests-prod.sh
View File

@ -17,10 +17,6 @@ download_material "${TASKCLUSTER_TMP_DIR}/ds"
export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH
if [ "${OS}" = "Darwin" ]; then
export DYLD_LIBRARY_PATH=$TESTS_BREW/lib/:$DYLD_LIBRARY_PATH
fi;
check_versions check_versions
run_prod_inference_tests "${bitrate}" run_prod_inference_tests "${bitrate}"

4
taskcluster/tc-cpp-ds-tests.sh
View File

@ -11,10 +11,6 @@ download_material "${TASKCLUSTER_TMP_DIR}/ds"
export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH
if [ "${OS}" = "Darwin" ]; then
export DYLD_LIBRARY_PATH=$TESTS_BREW/lib/:$DYLD_LIBRARY_PATH
fi;
check_versions check_versions
run_all_inference_tests run_all_inference_tests

4
taskcluster/tc-cpp_tflite-ds-tests.sh
View File

@ -16,10 +16,6 @@ download_material "${TASKCLUSTER_TMP_DIR}/ds"
export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH
if [ "${OS}" = "Darwin" ]; then
export DYLD_LIBRARY_PATH=$TESTS_BREW/lib/:$DYLD_LIBRARY_PATH
fi;
check_versions check_versions
run_all_inference_tests run_all_inference_tests

4
taskcluster/tc-cpp_tflite-tests-prod.sh
View File

@ -18,10 +18,6 @@ download_material "${TASKCLUSTER_TMP_DIR}/ds"
export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH
if [ "${OS}" = "Darwin" ]; then
export DYLD_LIBRARY_PATH=$TESTS_BREW/lib/:$DYLD_LIBRARY_PATH
fi;
check_versions check_versions
run_prodtflite_inference_tests "${bitrate}" run_prodtflite_inference_tests "${bitrate}"

4
taskcluster/tc-cpp_tflite_basic-ds-tests.sh
View File

@ -16,10 +16,6 @@ download_material "${TASKCLUSTER_TMP_DIR}/ds"
export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH export PATH=${TASKCLUSTER_TMP_DIR}/ds/:$PATH
if [ "${OS}" = "Darwin" ]; then
export DYLD_LIBRARY_PATH=$TESTS_BREW/lib/:$DYLD_LIBRARY_PATH
fi;
check_versions check_versions
run_tflite_basic_inference_tests run_tflite_basic_inference_tests

5
taskcluster/tf_darwin-amd64-opt.yml
View File

@ -5,12 +5,9 @@ build:
artifact_namespace: ${system.tensorflow.darwin_amd64.namespace} artifact_namespace: ${system.tensorflow.darwin_amd64.namespace}
generic: generic:
workerType: "ds-macos-heavy" workerType: "ds-macos-heavy"
system_config:
>
${tensorflow.packages_macos.brew}
scripts: scripts:
setup: "taskcluster/tf_tc-setup.sh" setup: "taskcluster/tf_tc-setup.sh"
build: "taskcluster/tf_tc-build.sh --osx" build: "taskcluster/tf_tc-build.sh --cpu"
package: "taskcluster/tf_tc-package.sh" package: "taskcluster/tf_tc-package.sh"
maxRunTime: 28800 maxRunTime: 28800
metadata: metadata:

15
taskcluster/tf_ios-arm64-opt.yml Normal file
View File

@ -0,0 +1,15 @@
build:
template_file: generic_tc_caching-darwin-opt-base.tyml
cache:
artifact_url: ${system.tensorflow.ios_arm64.url}
artifact_namespace: ${system.tensorflow.ios_arm64.namespace}
generic:
workerType: "ds-macos-heavy"
scripts:
setup: "taskcluster/tf_tc-setup.sh"
build: "taskcluster/tf_tc-build.sh --ios-arm64"
package: "taskcluster/tf_tc-package.sh"
maxRunTime: 28800
metadata:
name: "TensorFlow iOS ARM64 TFLite"
description: "Building TensorFlow for iOS ARM64, TFLite, optimized version"

15
taskcluster/tf_ios-x86_64-opt.yml Normal file
View File

@ -0,0 +1,15 @@
build:
template_file: generic_tc_caching-darwin-opt-base.tyml
cache:
artifact_url: ${system.tensorflow.ios_x86_64.url}
artifact_namespace: ${system.tensorflow.ios_x86_64.namespace}
generic:
workerType: "ds-macos-heavy"
scripts:
setup: "taskcluster/tf_tc-setup.sh"
build: "taskcluster/tf_tc-build.sh --ios-x86_64"
package: "taskcluster/tf_tc-package.sh"
maxRunTime: 28800
metadata:
name: "TensorFlow iOS x86_64 TFLite"
description: "Building TensorFlow for iOS x86_64, TFLite, optimized version"

2
taskcluster/tf_linux-amd64-cpu-opt.yml
View File

@ -8,7 +8,7 @@ build:
${tensorflow.packages_xenial.apt} && ${java.packages_xenial.apt} ${tensorflow.packages_xenial.apt} && ${java.packages_xenial.apt}
scripts: scripts:
setup: "taskcluster/tf_tc-setup.sh" setup: "taskcluster/tf_tc-setup.sh"
build: "taskcluster/tf_tc-build.sh" build: "taskcluster/tf_tc-build.sh --cpu"
package: "taskcluster/tf_tc-package.sh" package: "taskcluster/tf_tc-package.sh"
maxRunTime: 14400 maxRunTime: 14400
metadata: metadata:

44
taskcluster/tf_tc-brew.sh
View File

@ -1,44 +0,0 @@
#!/bin/bash
set -ex
if [ -z "${TASKCLUSTER_TASK_DIR}" ]; then
echo "No TASKCLUSTER_TASK_DIR, aborting."
exit 1
fi
LOCAL_BREW="${TASKCLUSTER_TASK_DIR}/homebrew"
export PATH=${LOCAL_BREW}/bin:$PATH
export HOMEBREW_LOGS="${TASKCLUSTER_TASK_DIR}/homebrew.logs/"
export HOMEBREW_CACHE="${TASKCLUSTER_TASK_DIR}/homebrew.cache/"
# Never fail on pre-existing homebrew/ directory
mkdir -p "${LOCAL_BREW}" || true
mkdir -p "${HOMEBREW_CACHE}" || true
# Make sure to verify there is a 'brew' binary there, otherwise install things.
if [ ! -x "${LOCAL_BREW}/bin/brew" ]; then
curl -L https://github.com/Homebrew/brew/tarball/2.2.17 | tar xz --strip 1 -C "${LOCAL_BREW}"
fi;
echo "local brew list (should be empty) ..."
brew list
echo "local brew prefix ..."
local_prefix=$(brew --prefix)
echo "${local_prefix}"
if [ "${LOCAL_BREW}" != "${local_prefix}" ]; then
echo "Weird state:"
echo "LOCAL_BREW=${LOCAL_BREW}"
echo "local_prefix=${local_prefix}"
exit 1
fi;
# coreutils, pyenv-virtualenv required for build of tensorflow
all_pkgs="coreutils pyenv-virtualenv"
for pkg in ${all_pkgs};
do
(brew list --versions ${pkg} && brew upgrade ${pkg}) || brew install ${pkg}
done;

111
taskcluster/tf_tc-build.sh
View File

@ -4,58 +4,6 @@ set -ex
source $(dirname $0)/tf_tc-vars.sh source $(dirname $0)/tf_tc-vars.sh
build_amd64=yes
build_gpu=no
build_android_arm=no
build_android_arm64=no
build_linux_arm=no
build_linux_arm64=no
if [ "$1" = "--gpu" ]; then
build_amd64=yes
build_gpu=yes
build_android_arm=no
build_android_arm64=no
build_linux_arm=no
build_linux_arm64=no
fi
if [ "$1" = "--arm" ]; then
build_amd64=yes
build_gpu=no
build_android_arm=no
build_android_arm64=no
build_linux_arm=yes
build_linux_arm64=no
fi
if [ "$1" = "--arm64" ]; then
build_amd64=yes
build_gpu=no
build_android_arm=no
build_android_arm64=no
build_linux_arm=no
build_linux_arm64=yes
fi
if [ "$1" = "--android-armv7" ]; then
build_amd64=no
build_gpu=no
build_android_arm=yes
build_android_arm64=no
build_linux_arm=no
build_linux_arm64=no
fi
if [ "$1" = "--android-arm64" ]; then
build_amd64=no
build_gpu=no
build_android_arm=no
build_android_arm64=yes
build_linux_arm=no
build_linux_arm64=no
fi
pushd ${DS_ROOT_TASK}/DeepSpeech/ds/tensorflow/ pushd ${DS_ROOT_TASK}/DeepSpeech/ds/tensorflow/
BAZEL_BUILD="bazel ${BAZEL_OUTPUT_USER_ROOT} build -s --explain bazel_monolithic_tf.log --verbose_explanations --experimental_strict_action_env --config=monolithic" BAZEL_BUILD="bazel ${BAZEL_OUTPUT_USER_ROOT} build -s --explain bazel_monolithic_tf.log --verbose_explanations --experimental_strict_action_env --config=monolithic"
@ -66,43 +14,32 @@ pushd ${DS_ROOT_TASK}/DeepSpeech/ds/tensorflow/
# Force toolchain sync (useful on macOS ?) # Force toolchain sync (useful on macOS ?)
bazel ${BAZEL_OUTPUT_USER_ROOT} sync --configure bazel ${BAZEL_OUTPUT_USER_ROOT} sync --configure
if [ "${build_amd64}" = "yes" ]; then case "$1" in
# Pure amd64 CPU-only build "--cpu")
if [ "${OS}" = "${TC_MSYS_VERSION}" -a "${build_gpu}" = "no" ]; then echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_OPT_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LIB_CPP_API} ${BUILD_TARGET_LITE_LIB}
echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_OPT_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LIB_CPP_API} ${BUILD_TARGET_LITE_LIB} ;;
elif [ "${build_gpu}" = "no" -a "${build_linux_arm}" = "no" -a "${build_linux_arm64}" = "no" ]; then "--gpu")
echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_OPT_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LIB_CPP_API} ${BUILD_TARGET_LITE_LIB} eval "export ${TF_CUDA_FLAGS}" && (echo "" | TF_NEED_CUDA=1 ./configure) && ${BAZEL_BUILD} -c opt ${BAZEL_CUDA_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BAZEL_OPT_FLAGS} ${BUILD_TARGET_LIB_CPP_API}
fi ;;
"--arm")
# Cross RPi3 CPU-only build echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ARM_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB}
if [ "${build_linux_arm}" = "yes" ]; then ;;
echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ARM_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB} "--arm64")
fi echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ARM64_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB}
;;
# Cross ARM64 Cortex-A53 build "--android-armv7")
if [ "${build_linux_arm64}" = "yes" ]; then
echo "" | TF_NEED_CUDA=0 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ARM64_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB}
fi
# Pure amd64 GPU-enabled build
if [ "${build_gpu}" = "yes" ]; then
eval "export ${TF_CUDA_FLAGS}" && (echo "" | TF_NEED_CUDA=1 ./configure) && ${BAZEL_BUILD} -c opt ${BAZEL_CUDA_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BAZEL_OPT_FLAGS} ${BUILD_TARGET_LIB_CPP_API}
fi
fi
if [ "${build_android_arm}" = "yes" ]; then
echo "" | TF_SET_ANDROID_WORKSPACE=1 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ANDROID_ARM_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB} echo "" | TF_SET_ANDROID_WORKSPACE=1 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ANDROID_ARM_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB}
fi; ;;
"--android-arm64")
if [ "${build_android_arm64}" = "yes" ]; then
echo "" | TF_SET_ANDROID_WORKSPACE=1 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ANDROID_ARM64_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB} echo "" | TF_SET_ANDROID_WORKSPACE=1 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_ANDROID_ARM64_FLAGS} ${BAZEL_EXTRA_FLAGS} ${BUILD_TARGET_LITE_LIB}
fi; ;;
"--ios-arm64")
if [ $? -ne 0 ]; then echo "" | TF_NEED_CUDA=0 TF_CONFIGURE_IOS=1 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_IOS_ARM64_FLAGS} ${BUILD_TARGET_LITE_LIB}
# There was a failure, just account for it. ;;
echo "Build failure, please check the output above. Exit code was: $?" "--ios-x86_64")
return 1 echo "" | TF_NEED_CUDA=0 TF_CONFIGURE_IOS=1 ./configure && ${BAZEL_BUILD} -c opt ${BAZEL_IOS_X86_64_FLAGS} ${BUILD_TARGET_LITE_LIB}
fi ;;
esac
bazel ${BAZEL_OUTPUT_USER_ROOT} shutdown bazel ${BAZEL_OUTPUT_USER_ROOT} shutdown
popd popd

2
taskcluster/tf_tc-vars.sh
View File

@ -171,6 +171,8 @@ BAZEL_ARM64_FLAGS="--config=rpi3-armv8 --config=rpi3-armv8_opt --copt=-DTFLITE_W
BAZEL_ANDROID_ARM_FLAGS="--config=android --config=android_arm --action_env ANDROID_NDK_API_LEVEL=21 --cxxopt=-std=c++14 --copt=-D_GLIBCXX_USE_C99 --copt=-DTFLITE_WITH_RUY_GEMV" BAZEL_ANDROID_ARM_FLAGS="--config=android --config=android_arm --action_env ANDROID_NDK_API_LEVEL=21 --cxxopt=-std=c++14 --copt=-D_GLIBCXX_USE_C99 --copt=-DTFLITE_WITH_RUY_GEMV"
BAZEL_ANDROID_ARM64_FLAGS="--config=android --config=android_arm64 --action_env ANDROID_NDK_API_LEVEL=21 --cxxopt=-std=c++14 --copt=-D_GLIBCXX_USE_C99 --copt=-DTFLITE_WITH_RUY_GEMV" BAZEL_ANDROID_ARM64_FLAGS="--config=android --config=android_arm64 --action_env ANDROID_NDK_API_LEVEL=21 --cxxopt=-std=c++14 --copt=-D_GLIBCXX_USE_C99 --copt=-DTFLITE_WITH_RUY_GEMV"
BAZEL_CUDA_FLAGS="--config=cuda" BAZEL_CUDA_FLAGS="--config=cuda"
BAZEL_IOS_ARM64_FLAGS="--config=ios_arm64 --define=runtime=tflite --copt=-DTFLITE_WITH_RUY_GEMV"
BAZEL_IOS_X86_64_FLAGS="--config=ios_x86_64 --define=runtime=tflite --copt=-DTFLITE_WITH_RUY_GEMV"
if [ "${OS}" = "${TC_MSYS_VERSION}" ]; then if [ "${OS}" = "${TC_MSYS_VERSION}" ]; then
# Somehow, even with Python being in the PATH, Bazel on windows struggles # Somehow, even with Python being in the PATH, Bazel on windows struggles

2
taskcluster/tf_win-amd64-cpu-opt.yml
View File

@ -8,7 +8,7 @@ build:
${tensorflow.packages_win.pacman} && ${tensorflow.packages_win.msys64} ${tensorflow.packages_win.pacman} && ${tensorflow.packages_win.msys64}
scripts: scripts:
setup: "taskcluster/tf_tc-setup.sh" setup: "taskcluster/tf_tc-setup.sh"
build: "taskcluster/tf_tc-build.sh" build: "taskcluster/tf_tc-build.sh --cpu"
package: "taskcluster/tf_tc-package.sh" package: "taskcluster/tf_tc-package.sh"
maxRunTime: 14400 maxRunTime: 14400
metadata: metadata:

2
taskcluster/win-opt-base.tyml
View File

@ -43,7 +43,7 @@ payload:
- .\msys64\usr\bin\bash.exe --login -cx "pacman -Syu --noconfirm" - .\msys64\usr\bin\bash.exe --login -cx "pacman -Syu --noconfirm"
- echo .\msys64\usr\bin\bash.exe --login -cxe " - echo .\msys64\usr\bin\bash.exe --login -cxe "
export LC_ALL=C && export LC_ALL=C &&
export PATH=\"/c/builds/tc-workdir/msys64/usr/bin:/c/Python36:/c/Program Files/Git/bin:/c/Program Files/7-Zip/:/c/Program Files (x86)/Windows Kits/10/bin/x64/:$PATH\" && export PATH=\"/c/builds/tc-workdir/msys64/usr/bin:/c/Python36:/c/Program Files/Git/bin:/c/Program Files/7-Zip/:/c/Program Files (x86)/Windows Kits/10/bin/x64/:/c/Program Files/dotnet/:$PATH\" &&
export TASKCLUSTER_ARTIFACTS=\"$USERPROFILE/public\" && export TASKCLUSTER_ARTIFACTS=\"$USERPROFILE/public\" &&
export TASKCLUSTER_TASK_DIR=\"/c/builds/tc-workdir/\" && export TASKCLUSTER_TASK_DIR=\"/c/builds/tc-workdir/\" &&
(mkdir $TASKCLUSTER_TASK_DIR || rm -fr $TASKCLUSTER_TASK_DIR/*) && cd $TASKCLUSTER_TASK_DIR && (mkdir $TASKCLUSTER_TASK_DIR || rm -fr $TASKCLUSTER_TASK_DIR/*) && cd $TASKCLUSTER_TASK_DIR &&

4
taskcluster/worker.cyml
View File

@ -4,8 +4,8 @@ taskcluster:
provisionerId: proj-deepspeech provisionerId: proj-deepspeech
workerType: ci workerType: ci
workerTypeKvm: kvm workerTypeKvm: kvm
workerTypeWin: win workerTypeWin: win-b
workerTypeCuda: win-gpu-b workerTypeCuda: win-gpu
dockerrpi3: dockerrpi3:
provisionerId: proj-deepspeech provisionerId: proj-deepspeech
workerType: ds-rpi3 workerType: ds-rpi3

2
tensorflow

@ -1 +1 @@
Subproject commit 518c1d04bf55d362bb11e973b8f5d0aa3e5bf44d Subproject commit 0854bb5188a3150a4d75a1c71ee610b0d45cfcb1

6
training/deepspeech_training/evaluate.py
View File

@ -50,7 +50,11 @@ def evaluate(test_csvs, create_model):
else: else:
scorer = None scorer = None
test_sets = [create_dataset([csv], batch_size=FLAGS.test_batch_size, train_phase=False) for csv in test_csvs] test_sets = [create_dataset([csv],
batch_size=FLAGS.test_batch_size,
train_phase=False,
reverse=FLAGS.reverse_test,
limit=FLAGS.limit_test) for csv in test_csvs]
iterator = tfv1.data.Iterator.from_structure(tfv1.data.get_output_types(test_sets[0]), iterator = tfv1.data.Iterator.from_structure(tfv1.data.get_output_types(test_sets[0]),
tfv1.data.get_output_shapes(test_sets[0]), tfv1.data.get_output_shapes(test_sets[0]),
output_classes=tfv1.data.get_output_classes(test_sets[0])) output_classes=tfv1.data.get_output_classes(test_sets[0]))

6
training/deepspeech_training/train.py
View File

@ -417,6 +417,8 @@ def train():
train_phase=True, train_phase=True,
exception_box=exception_box, exception_box=exception_box,
process_ahead=len(Config.available_devices) * FLAGS.train_batch_size * 2, process_ahead=len(Config.available_devices) * FLAGS.train_batch_size * 2,
reverse=FLAGS.reverse_train,
limit=FLAGS.limit_train,
buffering=FLAGS.read_buffer) buffering=FLAGS.read_buffer)
iterator = tfv1.data.Iterator.from_structure(tfv1.data.get_output_types(train_set), iterator = tfv1.data.Iterator.from_structure(tfv1.data.get_output_types(train_set),
@ -433,6 +435,8 @@ def train():
train_phase=False, train_phase=False,
exception_box=exception_box, exception_box=exception_box,
process_ahead=len(Config.available_devices) * FLAGS.dev_batch_size * 2, process_ahead=len(Config.available_devices) * FLAGS.dev_batch_size * 2,
reverse=FLAGS.reverse_dev,
limit=FLAGS.limit_dev,
buffering=FLAGS.read_buffer) for source in dev_sources] buffering=FLAGS.read_buffer) for source in dev_sources]
dev_init_ops = [iterator.make_initializer(dev_set) for dev_set in dev_sets] dev_init_ops = [iterator.make_initializer(dev_set) for dev_set in dev_sets]
@ -443,6 +447,8 @@ def train():
train_phase=False, train_phase=False,
exception_box=exception_box, exception_box=exception_box,
process_ahead=len(Config.available_devices) * FLAGS.dev_batch_size * 2, process_ahead=len(Config.available_devices) * FLAGS.dev_batch_size * 2,
reverse=FLAGS.reverse_dev,
limit=FLAGS.limit_dev,
buffering=FLAGS.read_buffer) for source in metrics_sources] buffering=FLAGS.read_buffer) for source in metrics_sources]
metrics_init_ops = [iterator.make_initializer(metrics_set) for metrics_set in metrics_sets] metrics_init_ops = [iterator.make_initializer(metrics_set) for metrics_set in metrics_sets]

9
training/deepspeech_training/util/augmentations.py
View File

@ -349,8 +349,13 @@ class Resample(SampleAugmentation):
audio = sample.audio audio = sample.audio
orig_len = len(audio) orig_len = len(audio)
audio = np.swapaxes(audio, 0, 1) audio = np.swapaxes(audio, 0, 1)
audio = resample(audio, sample.audio_format.rate, rate) if audio.shape[0] < 2:
audio = resample(audio, rate, sample.audio_format.rate) # since v0.8 librosa enforces a shape of (samples,) instead of (channels, samples) for mono samples
resampled = resample(audio[0], sample.audio_format.rate, rate)
audio[0] = resample(resampled, rate, sample.audio_format.rate)[:orig_len]
else:
audio = resample(audio, sample.audio_format.rate, rate)
audio = resample(audio, rate, sample.audio_format.rate)
audio = np.swapaxes(audio, 0, 1)[0:orig_len] audio = np.swapaxes(audio, 0, 1)[0:orig_len]
sample.audio = audio sample.audio = audio

8
training/deepspeech_training/util/feeding.py
View File

@ -90,6 +90,8 @@ def create_dataset(sources,
augmentations=None, augmentations=None,
cache_path=None, cache_path=None,
train_phase=False, train_phase=False,
reverse=False,
limit=0,
exception_box=None, exception_box=None,
process_ahead=None, process_ahead=None,
buffering=1 * MEGABYTE): buffering=1 * MEGABYTE):
@ -99,8 +101,10 @@ def create_dataset(sources,
epoch = epoch_counter['epoch'] epoch = epoch_counter['epoch']
if train_phase: if train_phase:
epoch_counter['epoch'] += 1 epoch_counter['epoch'] += 1
samples = samples_from_sources(sources, buffering=buffering, labeled=True) samples = samples_from_sources(sources, buffering=buffering, labeled=True, reverse=reverse)
num_samples = len(samples) num_samples = len(samples)
if limit > 0:
num_samples = min(limit, num_samples)
samples = apply_sample_augmentations(samples, samples = apply_sample_augmentations(samples,
augmentations, augmentations,
buffering=buffering, buffering=buffering,
@ -108,6 +112,8 @@ def create_dataset(sources,
clock=epoch / epochs, clock=epoch / epochs,
final_clock=(epoch + 1) / epochs) final_clock=(epoch + 1) / epochs)
for sample_index, sample in enumerate(samples): for sample_index, sample in enumerate(samples):
if sample_index >= num_samples:
break
clock = (epoch * num_samples + sample_index) / (epochs * num_samples) if train_phase and epochs > 0 else 0.0 clock = (epoch * num_samples + sample_index) / (epochs * num_samples) if train_phase and epochs > 0 else 0.0
transcript = text_to_char_array(sample.transcript, Config.alphabet, context=sample.sample_id) transcript = text_to_char_array(sample.transcript, Config.alphabet, context=sample.sample_id)
transcript = to_sparse_tuple(transcript) transcript = to_sparse_tuple(transcript)

10
training/deepspeech_training/util/flags.py
View File

@ -71,8 +71,14 @@ def create_flags():
# Sample limits # Sample limits
f.DEFINE_integer('limit_train', 0, 'maximum number of elements to use from train set - 0 means no limit') f.DEFINE_integer('limit_train', 0, 'maximum number of elements to use from train set - 0 means no limit')
f.DEFINE_integer('limit_dev', 0, 'maximum number of elements to use from validation set- 0 means no limit') f.DEFINE_integer('limit_dev', 0, 'maximum number of elements to use from validation set - 0 means no limit')
f.DEFINE_integer('limit_test', 0, 'maximum number of elements to use from test set- 0 means no limit') f.DEFINE_integer('limit_test', 0, 'maximum number of elements to use from test set - 0 means no limit')
# Sample order
f.DEFINE_boolean('reverse_train', False, 'if to reverse sample order of the train set')
f.DEFINE_boolean('reverse_dev', False, 'if to reverse sample order of the dev set')
f.DEFINE_boolean('reverse_test', False, 'if to reverse sample order of the test set')
# Checkpointing # Checkpointing

5
training/deepspeech_training/util/helpers.py
View File

@ -65,13 +65,14 @@ class Interleaved:
"""Collection that lazily combines sorted collections in an interleaving fashion. """Collection that lazily combines sorted collections in an interleaving fashion.
During iteration the next smallest element from all the sorted collections is always picked. During iteration the next smallest element from all the sorted collections is always picked.
The collections must support iter() and len().""" The collections must support iter() and len()."""
def __init__(self, *iterables, key=lambda obj: obj): def __init__(self, *iterables, key=lambda obj: obj, reverse=False):
self.iterables = iterables self.iterables = iterables
self.key = key self.key = key
self.reverse = reverse
self.len = sum(map(len, iterables)) self.len = sum(map(len, iterables))
def __iter__(self): def __iter__(self):
return heapq.merge(*self.iterables, key=self.key) return heapq.merge(*self.iterables, key=self.key, reverse=self.reverse)
def __len__(self): def __len__(self):
return self.len return self.len

121
training/deepspeech_training/util/sample_collections.py
View File

@ -6,8 +6,16 @@ import json
from pathlib import Path from pathlib import Path
from functools import partial from functools import partial
from .helpers import MEGABYTE, GIGABYTE, Interleaved from .helpers import KILOBYTE, MEGABYTE, GIGABYTE, Interleaved
from .audio import Sample, DEFAULT_FORMAT, AUDIO_TYPE_OPUS, SERIALIZABLE_AUDIO_TYPES, get_audio_type_from_extension from .audio import (
Sample,
DEFAULT_FORMAT,
AUDIO_TYPE_PCM,
AUDIO_TYPE_OPUS,
SERIALIZABLE_AUDIO_TYPES,
get_audio_type_from_extension,
write_wav
)
BIG_ENDIAN = 'big' BIG_ENDIAN = 'big'
INT_SIZE = 4 INT_SIZE = 4
@ -15,6 +23,7 @@ BIGINT_SIZE = 2 * INT_SIZE
MAGIC = b'SAMPLEDB' MAGIC = b'SAMPLEDB'
BUFFER_SIZE = 1 * MEGABYTE BUFFER_SIZE = 1 * MEGABYTE
REVERSE_BUFFER_SIZE = 16 * KILOBYTE
CACHE_SIZE = 1 * GIGABYTE CACHE_SIZE = 1 * GIGABYTE
SCHEMA_KEY = 'schema' SCHEMA_KEY = 'schema'
@ -181,14 +190,19 @@ class DirectSDBWriter:
class SDB: # pylint: disable=too-many-instance-attributes class SDB: # pylint: disable=too-many-instance-attributes
"""Sample collection reader for reading a Sample DB (SDB) file""" """Sample collection reader for reading a Sample DB (SDB) file"""
def __init__(self, sdb_filename, buffering=BUFFER_SIZE, id_prefix=None, labeled=True): def __init__(self,
sdb_filename,
buffering=BUFFER_SIZE,
id_prefix=None,
labeled=True,
reverse=False):
""" """
Parameters Parameters
---------- ----------
sdb_filename : str sdb_filename : str
Path to the SDB file to read samples from Path to the SDB file to read samples from
buffering : int buffering : int
Read-buffer size to use while reading the SDB file Read-ahead buffer size to use while reading the SDB file in normal order. Fixed to 16kB if in reverse-mode.
id_prefix : str id_prefix : str
Prefix for IDs of read samples - defaults to sdb_filename Prefix for IDs of read samples - defaults to sdb_filename
labeled : bool or None labeled : bool or None
@ -199,7 +213,7 @@ class SDB: # pylint: disable=too-many-instance-attributes
""" """
self.sdb_filename = sdb_filename self.sdb_filename = sdb_filename
self.id_prefix = sdb_filename if id_prefix is None else id_prefix self.id_prefix = sdb_filename if id_prefix is None else id_prefix
self.sdb_file = open(sdb_filename, 'rb', buffering=buffering) self.sdb_file = open(sdb_filename, 'rb', buffering=REVERSE_BUFFER_SIZE if reverse else buffering)
self.offsets = [] self.offsets = []
if self.sdb_file.read(len(MAGIC)) != MAGIC: if self.sdb_file.read(len(MAGIC)) != MAGIC:
raise RuntimeError('No Sample Database') raise RuntimeError('No Sample Database')
@ -229,6 +243,8 @@ class SDB: # pylint: disable=too-many-instance-attributes
num_samples = self.read_big_int() num_samples = self.read_big_int()
for _ in range(num_samples): for _ in range(num_samples):
self.offsets.append(self.read_big_int()) self.offsets.append(self.read_big_int())
if reverse:
self.offsets.reverse()
def read_int(self): def read_int(self):
return int.from_bytes(self.sdb_file.read(INT_SIZE), BIG_ENDIAN) return int.from_bytes(self.sdb_file.read(INT_SIZE), BIG_ENDIAN)
@ -297,9 +313,73 @@ class SDB: # pylint: disable=too-many-instance-attributes
self.close() self.close()
class CSVWriter: # pylint: disable=too-many-instance-attributes
"""Sample collection writer for writing a CSV data-set and all its referenced WAV samples"""
def __init__(self,
csv_filename,
absolute_paths=False,
labeled=True):
"""
Parameters
----------
csv_filename : str
Path to the CSV file to write.
Will create a directory (CSV-filename without extension) next to it and fail if it already exists.
absolute_paths : bool
If paths in CSV file should be absolute instead of relative to the CSV file's parent directory.
labeled : bool or None
If True: Writes labeled samples (util.sample_collections.LabeledSample) only.
If False: Ignores transcripts (if available) and writes (unlabeled) util.audio.Sample instances.
"""
self.csv_filename = Path(csv_filename)
self.csv_base_dir = self.csv_filename.parent.resolve().absolute()
self.set_name = self.csv_filename.stem
self.csv_dir = self.csv_base_dir / self.set_name
if self.csv_dir.exists():
raise RuntimeError('"{}" already existing'.format(self.csv_dir))
os.mkdir(str(self.csv_dir))
self.absolute_paths = absolute_paths
fieldnames = ['wav_filename', 'wav_filesize']
self.labeled = labeled
if labeled:
fieldnames.append('transcript')
self.csv_file = open(csv_filename, 'w', encoding='utf-8', newline='')
self.csv_writer = csv.DictWriter(self.csv_file, fieldnames=fieldnames)
self.csv_writer.writeheader()
self.counter = 0
def __enter__(self):
return self
def add(self, sample):
sample_filename = self.csv_dir / 'sample{0:08d}.wav'.format(self.counter)
self.counter += 1
sample.change_audio_type(AUDIO_TYPE_PCM)
write_wav(str(sample_filename), sample.audio, audio_format=sample.audio_format)
sample.sample_id = str(sample_filename.relative_to(self.csv_base_dir))
row = {
'wav_filename': str(sample_filename.absolute()) if self.absolute_paths else sample.sample_id,
'wav_filesize': sample_filename.stat().st_size
}
if self.labeled:
row['transcript'] = sample.transcript
self.csv_writer.writerow(row)
return sample.sample_id
def close(self):
if self.csv_file:
self.csv_file.close()
def __len__(self):
return self.counter
def __exit__(self, exc_type, exc_val, exc_tb):
self.close()
class SampleList: class SampleList:
"""Sample collection base class with samples loaded from a list of in-memory paths.""" """Sample collection base class with samples loaded from a list of in-memory paths."""
def __init__(self, samples, labeled=True): def __init__(self, samples, labeled=True, reverse=False):
""" """
Parameters Parameters
---------- ----------
@ -308,10 +388,12 @@ class SampleList:
labeled : bool or None labeled : bool or None
If True: Reads LabeledSample instances. If True: Reads LabeledSample instances.
If False: Ignores transcripts (if available) and reads (unlabeled) util.audio.Sample instances. If False: Ignores transcripts (if available) and reads (unlabeled) util.audio.Sample instances.
reverse : bool
If the order of the samples should be reversed
""" """
self.labeled = labeled self.labeled = labeled
self.samples = list(samples) self.samples = list(samples)
self.samples.sort(key=lambda r: r[1]) self.samples.sort(key=lambda r: r[1], reverse=reverse)
def __getitem__(self, i): def __getitem__(self, i):
sample_spec = self.samples[i] sample_spec = self.samples[i]
@ -324,7 +406,7 @@ class SampleList:
class CSV(SampleList): class CSV(SampleList):
"""Sample collection reader for reading a DeepSpeech CSV file """Sample collection reader for reading a DeepSpeech CSV file
Automatically orders samples by CSV column wav_filesize (if available).""" Automatically orders samples by CSV column wav_filesize (if available)."""
def __init__(self, csv_filename, labeled=None): def __init__(self, csv_filename, labeled=None, reverse=False):
""" """
Parameters Parameters
---------- ----------
@ -335,6 +417,8 @@ class CSV(SampleList):
If False: Ignores transcripts (if available) and reads (unlabeled) util.audio.Sample instances. If False: Ignores transcripts (if available) and reads (unlabeled) util.audio.Sample instances.
If None: Automatically determines if CSV file has a transcript column If None: Automatically determines if CSV file has a transcript column
(reading util.sample_collections.LabeledSample instances) or not (reading util.audio.Sample instances). (reading util.sample_collections.LabeledSample instances) or not (reading util.audio.Sample instances).
reverse : bool
If the order of the samples should be reversed
""" """
rows = [] rows = []
csv_dir = Path(csv_filename).parent csv_dir = Path(csv_filename).parent
@ -355,10 +439,10 @@ class CSV(SampleList):
rows.append((wav_filename, wav_filesize, row['transcript'])) rows.append((wav_filename, wav_filesize, row['transcript']))
else: else:
rows.append((wav_filename, wav_filesize)) rows.append((wav_filename, wav_filesize))
super(CSV, self).__init__(rows, labeled=labeled) super(CSV, self).__init__(rows, labeled=labeled, reverse=reverse)
def samples_from_source(sample_source, buffering=BUFFER_SIZE, labeled=None): def samples_from_source(sample_source, buffering=BUFFER_SIZE, labeled=None, reverse=False):
""" """
Loads samples from a sample source file. Loads samples from a sample source file.
@ -373,6 +457,8 @@ def samples_from_source(sample_source, buffering=BUFFER_SIZE, labeled=None):
If False: Ignores transcripts (if available) and reads (unlabeled) util.audio.Sample instances. If False: Ignores transcripts (if available) and reads (unlabeled) util.audio.Sample instances.
If None: Automatically determines if source provides transcripts If None: Automatically determines if source provides transcripts
(reading util.sample_collections.LabeledSample instances) or not (reading util.audio.Sample instances). (reading util.sample_collections.LabeledSample instances) or not (reading util.audio.Sample instances).
reverse : bool
If the order of the samples should be reversed
Returns Returns
------- -------
@ -380,13 +466,13 @@ def samples_from_source(sample_source, buffering=BUFFER_SIZE, labeled=None):
""" """
ext = os.path.splitext(sample_source)[1].lower() ext = os.path.splitext(sample_source)[1].lower()
if ext == '.sdb': if ext == '.sdb':
return SDB(sample_source, buffering=buffering, labeled=labeled) return SDB(sample_source, buffering=buffering, labeled=labeled, reverse=reverse)
if ext == '.csv': if ext == '.csv':
return CSV(sample_source, labeled=labeled) return CSV(sample_source, labeled=labeled, reverse=reverse)
raise ValueError('Unknown file type: "{}"'.format(ext)) raise ValueError('Unknown file type: "{}"'.format(ext))
def samples_from_sources(sample_sources, buffering=BUFFER_SIZE, labeled=None): def samples_from_sources(sample_sources, buffering=BUFFER_SIZE, labeled=None, reverse=False):
""" """
Loads and combines samples from a list of source files. Sources are combined in an interleaving way to Loads and combines samples from a list of source files. Sources are combined in an interleaving way to
keep default sample order from shortest to longest. keep default sample order from shortest to longest.
@ -402,6 +488,8 @@ def samples_from_sources(sample_sources, buffering=BUFFER_SIZE, labeled=None):
If False: Ignores transcripts (if available) and always reads (unlabeled) util.audio.Sample instances. If False: Ignores transcripts (if available) and always reads (unlabeled) util.audio.Sample instances.
If None: Reads util.sample_collections.LabeledSample instances from sources with transcripts and If None: Reads util.sample_collections.LabeledSample instances from sources with transcripts and
util.audio.Sample instances from sources with no transcripts. util.audio.Sample instances from sources with no transcripts.
reverse : bool
If the order of the samples should be reversed
Returns Returns
------- -------
@ -411,6 +499,7 @@ def samples_from_sources(sample_sources, buffering=BUFFER_SIZE, labeled=None):
if len(sample_sources) == 0: if len(sample_sources) == 0:
raise ValueError('No files') raise ValueError('No files')
if len(sample_sources) == 1: if len(sample_sources) == 1:
return samples_from_source(sample_sources[0], buffering=buffering, labeled=labeled) return samples_from_source(sample_sources[0], buffering=buffering, labeled=labeled, reverse=reverse)
cols = list(map(partial(samples_from_source, buffering=buffering, labeled=labeled), sample_sources)) cols = [samples_from_source(source, buffering=buffering, labeled=labeled, reverse=reverse)
return Interleaved(*cols, key=lambda s: s.duration) for source in sample_sources]
return Interleaved(*cols, key=lambda s: s.duration, reverse=reverse)

22
training/deepspeech_training/util/text.py
View File

@ -9,16 +9,20 @@ def text_to_char_array(transcript, alphabet, context=''):
integers and return a numpy array representing the processed string. integers and return a numpy array representing the processed string.
Use a string in `context` for adding text to raised exceptions. Use a string in `context` for adding text to raised exceptions.
""" """
try: if not alphabet.CanEncode(transcript):
transcript = alphabet.Encode(transcript)
if len(transcript) == 0:
raise ValueError('While processing {}: Found an empty transcript! '
'You must include a transcript for all training data.'
.format(context))
return transcript
except KeyError as e:
# Provide the row context (especially wav_filename) for alphabet errors # Provide the row context (especially wav_filename) for alphabet errors
raise ValueError('While processing: {}\n{}'.format(context, e)) raise ValueError(
'Alphabet cannot encode transcript "{}" while processing sample "{}", '
'check that your alphabet contains all characters in the training corpus. '
'Missing characters are: {}.'
.format(transcript, context, list(ch for ch in transcript if not alphabet.CanEncodeSingle(ch))))
encoded = alphabet.Encode(transcript)
if len(encoded) == 0:
raise ValueError('While processing {}: Found an empty transcript! '
'You must include a transcript for all training data.'
.format(context))
return encoded
# The following code is from: http://hetland.org/coding/python/levenshtein.py # The following code is from: http://hetland.org/coding/python/levenshtein.py