Merge pull request #2591 from mozilla/revert-2548-net-streams

Revert "Multi-stream support .NET"

doc/DotNet-API.rst

@@ -15,13 +15,6 @@ DeepSpeech Class
    :project: deepspeech-dotnet
    :members:
 
-DeepSpeechStream Class
-----------------
-
-.. doxygenclass:: DeepSpeechClient::DeepSpeechStream
-   :project: deepspeech-dotnet
-   :members:
-
 ErrorCodes
 ----------
 

examples/net_framework/DeepSpeechWPF/App.xaml.cs

@@ -18,20 +18,20 @@ namespace DeepSpeechWPF
 
             const int BEAM_WIDTH = 500;
 
+            //Register instance of DeepSpeech
+            DeepSpeechClient.DeepSpeech deepSpeechClient = new DeepSpeechClient.DeepSpeech();
             try
             {
-                //Register instance of DeepSpeech
+                deepSpeechClient.CreateModel("output_graph.pbmm", BEAM_WIDTH);
-                DeepSpeechClient.DeepSpeech deepSpeechClient =
-                    new DeepSpeechClient.DeepSpeech("output_graph.pbmm", BEAM_WIDTH);
-
-                SimpleIoc.Default.Register<IDeepSpeech>(() => deepSpeechClient);
-                SimpleIoc.Default.Register<MainWindowViewModel>();
             }
             catch (System.Exception ex)
             {
                 MessageBox.Show(ex.Message);
                 Current.Shutdown();
             }
+            
+            SimpleIoc.Default.Register<IDeepSpeech>(() => deepSpeechClient);
+            SimpleIoc.Default.Register<MainWindowViewModel>();
         }
 
         protected override void OnExit(ExitEventArgs e)

examples/net_framework/DeepSpeechWPF/ViewModels/MainWindowViewModel.cs

@@ -4,7 +4,6 @@ using CSCore.CoreAudioAPI;
 using CSCore.SoundIn;
 using CSCore.Streams;
 using DeepSpeechClient.Interfaces;
-using DeepSpeechClient.Models;
 using GalaSoft.MvvmLight.CommandWpf;
 using Microsoft.Win32;
 using System;
@@ -59,12 +58,6 @@ namespace DeepSpeech.WPF.ViewModels
         #endregion
 
         #region Streaming
-
-        /// <summary>
-        /// Stream used to feed data into the acoustic model.
-        /// </summary>
-        private DeepSpeechStream _sttStream;
-
         /// <summary>
         /// Records the audio of the selected device.
         /// </summary>
@@ -315,7 +308,7 @@ namespace DeepSpeech.WPF.ViewModels
                 if (_bufferQueue.TryDequeue(out short[] buffer))
                 {
                     StreamingIsBusy = true;
-                    _sttClient.FeedAudioContent(_sttStream, buffer, Convert.ToUInt32(buffer.Length));
+                    _sttClient.FeedAudioContent(buffer, Convert.ToUInt32(buffer.Length));
                     StreamingIsBusy = false;
                 }
             }
@@ -393,7 +386,7 @@ namespace DeepSpeech.WPF.ViewModels
             {
                 await Task.Delay(90);
             }
-            Transcription = _sttClient.FinishStream(_sttStream);
+            Transcription = _sttClient.FinishStream();
             EnableStartRecord = true;
         }
 
@@ -402,7 +395,7 @@ namespace DeepSpeech.WPF.ViewModels
         /// </summary>
         private void StartRecording()
         {
-            _sttStream =_sttClient.CreateStream();
+            _sttClient.CreateStream();
             _audioCapture.Start();
             EnableStartRecord = false;
             EnableStopRecord = true;

native_client/dotnet/DeepSpeechClient/DeepSpeech.cs

@@ -4,7 +4,6 @@ using DeepSpeechClient.Extensions;
 using System;
 using System.IO;
 using DeepSpeechClient.Enums;
-using DeepSpeechClient.Models;
 
 namespace DeepSpeechClient
 {
@@ -14,16 +13,14 @@ namespace DeepSpeechClient
     public class DeepSpeech : IDeepSpeech
     {
         private unsafe IntPtr** _modelStatePP;
-        
+        private unsafe IntPtr** _streamingStatePP;
-        /// <summary>
+
-        /// Initializes a new instance of <see cref="DeepSpeech"/> class and creates a new acoustic model.
+
-        /// </summary>
+
-        /// <param name="aModelPath">The path to the frozen model graph.</param>
+
-        /// <param name="aBeamWidth">The beam width used by the decoder. A larger beam width generates better results at the cost of decoding time.</param>
+        public DeepSpeech()
-        /// <exception cref="ArgumentException">Thrown when the native binary failed to create the model.</exception>
-        public DeepSpeech(string aModelPath, uint aBeamWidth)
         {
-            CreateModel(aModelPath, aBeamWidth);
+
         }
 
         #region IDeepSpeech
@@ -34,7 +31,7 @@ namespace DeepSpeechClient
         /// <param name="aModelPath">The path to the frozen model graph.</param>
         /// <param name="aBeamWidth">The beam width used by the decoder. A larger beam width generates better results at the cost of decoding time.</param>
         /// <exception cref="ArgumentException">Thrown when the native binary failed to create the model.</exception>
-        private unsafe void CreateModel(string aModelPath,
+        public unsafe void CreateModel(string aModelPath,
             uint aBeamWidth)
         {
             string exceptionMessage = null;
@@ -121,19 +118,10 @@ namespace DeepSpeechClient
         /// <param name="aLMAlpha">The alpha hyperparameter of the CTC decoder. Language Model weight.</param>
         /// <param name="aLMBeta">The beta hyperparameter of the CTC decoder. Word insertion weight.</param>
         /// <exception cref="ArgumentException">Thrown when the native binary failed to enable decoding with a language model.</exception>
-        /// <exception cref="FileNotFoundException">Thrown when cannot find the language model or trie file.</exception>
         public unsafe void EnableDecoderWithLM(string aLMPath, string aTriePath,
             float aLMAlpha, float aLMBeta)
         {
             string exceptionMessage = null;
-            if (string.IsNullOrWhiteSpace(aLMPath))
-            {
-                exceptionMessage = "Path to the language model file cannot be empty.";
-            }
-            if (!File.Exists(aLMPath))
-            {
-                exceptionMessage = $"Cannot find the language model file: {aLMPath}";
-            }
             if (string.IsNullOrWhiteSpace(aTriePath))
             {
                 exceptionMessage = "Path to the trie file cannot be empty.";
@@ -159,41 +147,37 @@ namespace DeepSpeechClient
         /// <summary>
         /// Feeds audio samples to an ongoing streaming inference.
         /// </summary>
-        /// <param name="stream">Instance of the stream to feed the data.</param>
         /// <param name="aBuffer">An array of 16-bit, mono raw audio samples at the appropriate sample rate (matching what the model was trained on).</param>
-        public unsafe void FeedAudioContent(DeepSpeechStream stream, short[] aBuffer, uint aBufferSize)
+        public unsafe void FeedAudioContent(short[] aBuffer, uint aBufferSize)
         {
-            NativeImp.DS_FeedAudioContent(stream.GetNativePointer(), aBuffer, aBufferSize);
+            NativeImp.DS_FeedAudioContent(_streamingStatePP, aBuffer, aBufferSize);
         }
 
         /// <summary>
         /// Closes the ongoing streaming inference, returns the STT result over the whole audio signal.
         /// </summary>
-        /// <param name="stream">Instance of the stream to finish.</param>
+        /// <returns>The STT result. The user is responsible for freeing the string.</returns>
-        /// <returns>The STT result.</returns>
+        public unsafe string FinishStream()
-        public unsafe string FinishStream(DeepSpeechStream stream)
         {
-            return NativeImp.DS_FinishStream(stream.GetNativePointer()).PtrToString();
+            return NativeImp.DS_FinishStream(_streamingStatePP).PtrToString();
         }
 
         /// <summary>
         /// Closes the ongoing streaming inference, returns the STT result over the whole audio signal.
         /// </summary>
-        /// <param name="stream">Instance of the stream to finish.</param>
+        /// <returns>The extended metadata. The user is responsible for freeing the struct.</returns>
-        /// <returns>The extended metadata result.</returns>
+        public unsafe Models.Metadata FinishStreamWithMetadata()
-        public unsafe Metadata FinishStreamWithMetadata(DeepSpeechStream stream)
         {
-            return NativeImp.DS_FinishStreamWithMetadata(stream.GetNativePointer()).PtrToMetadata();
+            return NativeImp.DS_FinishStreamWithMetadata(_streamingStatePP).PtrToMetadata();
         }
 
         /// <summary>
         /// Computes the intermediate decoding of an ongoing streaming inference.
         /// </summary>
-        /// <param name="stream">Instance of the stream to decode.</param>
+        /// <returns>The STT intermediate result. The user is responsible for freeing the string.</returns>
-        /// <returns>The STT intermediate result.</returns>
+        public unsafe string IntermediateDecode()
-        public unsafe string IntermediateDecode(DeepSpeechStream stream)
         {
-            return NativeImp.DS_IntermediateDecode(stream.GetNativePointer());
+            return NativeImp.DS_IntermediateDecode(_streamingStatePP);
         }
 
         /// <summary>
@@ -207,12 +191,11 @@ namespace DeepSpeechClient
         /// <summary>
         /// Creates a new streaming inference state.
         /// </summary>
-        public unsafe DeepSpeechStream CreateStream()
+        /// <exception cref="ArgumentException">Thrown when the native binary failed to initialize the streaming mode.</exception>
+        public unsafe void CreateStream()
         {
-            IntPtr** streamingStatePointer = null;
+            var resultCode = NativeImp.DS_CreateStream(_modelStatePP, ref _streamingStatePP);
-            var resultCode = NativeImp.DS_CreateStream(_modelStatePP, ref streamingStatePointer);
             EvaluateResultCode(resultCode);
-            return new DeepSpeechStream(streamingStatePointer);
         }
 
         /// <summary>
@@ -220,10 +203,25 @@ namespace DeepSpeechClient
         /// This can be used if you no longer need the result of an ongoing streaming
         /// inference and don't want to perform a costly decode operation.
         /// </summary>
-        public unsafe void FreeStream(DeepSpeechStream stream)
+        public unsafe void FreeStream()
         {
-            NativeImp.DS_FreeStream(stream.GetNativePointer());
+            NativeImp.DS_FreeStream(ref _streamingStatePP);
-            stream.Dispose();
+        }
+
+        /// <summary>
+        /// Free a DeepSpeech allocated string
+        /// </summary>
+        public unsafe void FreeString(IntPtr intPtr)
+        {
+            NativeImp.DS_FreeString(intPtr);
+        }
+
+        /// <summary>
+        /// Free a DeepSpeech allocated Metadata struct
+        /// </summary>
+        public unsafe void FreeMetadata(IntPtr intPtr)
+        {
+            NativeImp.DS_FreeMetadata(intPtr);
         }
 
         /// <summary>
@@ -231,7 +229,7 @@ namespace DeepSpeechClient
         /// </summary>
         /// <param name="aBuffer">A 16-bit, mono raw audio signal at the appropriate sample rate (matching what the model was trained on).</param>
         /// <param name="aBufferSize">The number of samples in the audio signal.</param>
-        /// <returns>The STT result. Returns NULL on error.</returns>
+        /// <returns>The STT result. The user is responsible for freeing the string.  Returns NULL on error.</returns>
         public unsafe string SpeechToText(short[] aBuffer, uint aBufferSize)
         {
             return NativeImp.DS_SpeechToText(_modelStatePP, aBuffer, aBufferSize).PtrToString();
@@ -242,8 +240,8 @@ namespace DeepSpeechClient
         /// </summary>
         /// <param name="aBuffer">A 16-bit, mono raw audio signal at the appropriate sample rate (matching what the model was trained on).</param>
         /// <param name="aBufferSize">The number of samples in the audio signal.</param>
-        /// <returns>The extended metadata. Returns NULL on error.</returns>
+        /// <returns>The extended metadata. The user is responsible for freeing the struct.  Returns NULL on error.</returns>
-        public unsafe Metadata SpeechToTextWithMetadata(short[] aBuffer, uint aBufferSize)
+        public unsafe Models.Metadata SpeechToTextWithMetadata(short[] aBuffer, uint aBufferSize)
         {
             return NativeImp.DS_SpeechToTextWithMetadata(_modelStatePP, aBuffer, aBufferSize).PtrToMetadata();
         }

native_client/dotnet/DeepSpeechClient/DeepSpeechClient.csproj

@@ -48,7 +48,6 @@
     <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\MetadataItem.cs" />
     <Compile Include="NativeImp.cs" />

native_client/dotnet/DeepSpeechClient/Interfaces/IDeepSpeech.cs

@@ -1,11 +1,10 @@
 using DeepSpeechClient.Models;
 using System;
-using System.IO;
 
 namespace DeepSpeechClient.Interfaces
 {
     /// <summary>
-    /// Client interface of the Mozilla's DeepSpeech implementation.
+    /// Client interface of the Mozilla's deepspeech implementation.
     /// </summary>
     public interface IDeepSpeech : IDisposable
     {
@@ -14,6 +13,15 @@ namespace DeepSpeechClient.Interfaces
         /// </summary>
         void PrintVersions();
 
+        /// <summary>
+        /// Create an object providing an interface to a trained DeepSpeech model.
+        /// </summary>
+        /// <param name="aModelPath">The path to the frozen model graph.</param>
+        /// <param name="aBeamWidth">The beam width used by the decoder. A larger beam width generates better results at the cost of decoding time.</param>
+        /// <exception cref="ArgumentException">Thrown when the native binary failed to create the model.</exception>
+        unsafe void CreateModel(string aModelPath,
+                   uint aBeamWidth);
+
         /// <summary>
         /// Return the sample rate expected by the model.
         /// </summary>
@@ -28,7 +36,6 @@ namespace DeepSpeechClient.Interfaces
         /// <param name="aLMAlpha">The alpha hyperparameter of the CTC decoder. Language Model weight.</param>
         /// <param name="aLMBeta">The beta hyperparameter of the CTC decoder. Word insertion weight.</param>
         /// <exception cref="ArgumentException">Thrown when the native binary failed to enable decoding with a language model.</exception>
-        /// <exception cref="FileNotFoundException">Thrown when cannot find the language model or trie file.</exception>
         unsafe void EnableDecoderWithLM(string aLMPath,
                   string aTriePath,
                   float aLMAlpha,
@@ -39,7 +46,7 @@ namespace DeepSpeechClient.Interfaces
         /// </summary>
         /// <param name="aBuffer">A 16-bit, mono raw audio signal at the appropriate sample rate (matching what the model was trained on).</param>
         /// <param name="aBufferSize">The number of samples in the audio signal.</param>
-        /// <returns>The STT result. Returns NULL on error.</returns>
+        /// <returns>The STT result. The user is responsible for freeing the string.  Returns NULL on error.</returns>
         unsafe string SpeechToText(short[] aBuffer,
                 uint aBufferSize);
 
@@ -48,7 +55,7 @@ namespace DeepSpeechClient.Interfaces
         /// </summary>
         /// <param name="aBuffer">A 16-bit, mono raw audio signal at the appropriate sample rate (matching what the model was trained on).</param>
         /// <param name="aBufferSize">The number of samples in the audio signal.</param>
-        /// <returns>The extended metadata. Returns NULL on error.</returns>
+        /// <returns>The extended metadata result. The user is responsible for freeing the struct.  Returns NULL on error.</returns>
         unsafe Metadata SpeechToTextWithMetadata(short[] aBuffer,
                 uint aBufferSize);
 
@@ -57,39 +64,46 @@ namespace DeepSpeechClient.Interfaces
         /// This can be used if you no longer need the result of an ongoing streaming
         /// inference and don't want to perform a costly decode operation.
         /// </summary>
-        unsafe void FreeStream(DeepSpeechStream stream);
+        unsafe void FreeStream();
+
+        /// <summary>
+        /// Free a DeepSpeech allocated string
+        /// </summary>
+        unsafe void FreeString(IntPtr intPtr);
+
+        /// <summary>
+        /// Free a DeepSpeech allocated Metadata struct
+        /// </summary>
+        unsafe void FreeMetadata(IntPtr intPtr);
 
         /// <summary>
         /// Creates a new streaming inference state.
         /// </summary>
-        unsafe DeepSpeechStream CreateStream();
+        /// <exception cref="ArgumentException">Thrown when the native binary failed to initialize the streaming mode.</exception>
+        unsafe void CreateStream();
 
         /// <summary>
         /// Feeds audio samples to an ongoing streaming inference.
         /// </summary>
-        /// <param name="stream">Instance of the stream to feed the data.</param>
         /// <param name="aBuffer">An array of 16-bit, mono raw audio samples at the appropriate sample rate (matching what the model was trained on).</param>
-        unsafe void FeedAudioContent(DeepSpeechStream stream, short[] aBuffer, uint aBufferSize);
+        unsafe void FeedAudioContent(short[] aBuffer, uint aBufferSize);
 
         /// <summary>
         /// Computes the intermediate decoding of an ongoing streaming inference.
         /// </summary>
-        /// <param name="stream">Instance of the stream to decode.</param>
+        /// <returns>The STT intermediate result. The user is responsible for freeing the string.</returns>
-        /// <returns>The STT intermediate result.</returns>
+        unsafe string IntermediateDecode();
-        unsafe string IntermediateDecode(DeepSpeechStream stream);
 
         /// <summary>
         /// Closes the ongoing streaming inference, returns the STT result over the whole audio signal.
         /// </summary>
-        /// <param name="stream">Instance of the stream to finish.</param>
+        /// <returns>The STT result. The user is responsible for freeing the string.</returns>
-        /// <returns>The STT result.</returns>
+        unsafe string FinishStream();
-        unsafe string FinishStream(DeepSpeechStream stream);
 
         /// <summary>
         /// Closes the ongoing streaming inference, returns the STT result over the whole audio signal.
         /// </summary>
-        /// <param name="stream">Instance of the stream to finish.</param>
+        /// <returns>The extended metadata result. The user is responsible for freeing the struct.</returns>
-        /// <returns>The extended metadata result.</returns>
+        unsafe Metadata FinishStreamWithMetadata();
-        unsafe Metadata FinishStreamWithMetadata(DeepSpeechStream stream);
     }
 }

native_client/dotnet/DeepSpeechClient/Models/DeepSpeechStream.cs

@@ -1,35 +0,0 @@
-using System;
-
-namespace DeepSpeechClient.Models
-{
-    /// <summary>
-    /// Wrapper of the pointer used for the decoding stream.
-    /// </summary>
-    public class DeepSpeechStream : IDisposable
-    {
-        private unsafe IntPtr** _streamingStatePp;
-
-        /// <summary>
-        /// Initializes a new instance of <see cref="DeepSpeechStream"/>.
-        /// </summary>
-        /// <param name="streamingStatePP">Native pointer of the native stream.</param>
-        public unsafe DeepSpeechStream(IntPtr** streamingStatePP)
-        {
-            _streamingStatePp = streamingStatePP;
-        }
-
-        /// <summary>
-        /// Gets the native pointer.
-        /// </summary>
-        /// <exception cref="InvalidOperationException">Thrown when the stream has been disposed or not yet initialized.</exception>
-        /// <returns>Native pointer of the stream.</returns>
-        internal unsafe IntPtr** GetNativePointer()
-        {
-            if (_streamingStatePp == null)
-                throw new InvalidOperationException("Cannot use a disposed or uninitialized stream.");
-            return _streamingStatePp;
-        }
-
-        public unsafe void Dispose() => _streamingStatePp = null;
-    }
-}

native_client/dotnet/DeepSpeechClient/NativeImp.cs

@@ -48,7 +48,7 @@ namespace DeepSpeechClient
                ref IntPtr** retval);
 
         [DllImport("libdeepspeech.so", CallingConvention = CallingConvention.Cdecl)]
-        internal static unsafe extern void DS_FreeStream(IntPtr** aSctx);
+        internal static unsafe extern void DS_FreeStream(ref IntPtr** aSctx);
 
         [DllImport("libdeepspeech.so", CallingConvention = CallingConvention.Cdecl)]
         internal static unsafe extern void DS_FreeMetadata(IntPtr metadata);

native_client/dotnet/DeepSpeechConsole/Program.cs

@@ -53,13 +53,16 @@ namespace CSharpExamples
             const float LM_BETA = 1.85f;
 
             Stopwatch stopwatch = new Stopwatch();
-            try
+
+            using (IDeepSpeech sttClient = new DeepSpeech())
             {
-                Console.WriteLine("Loading model...");
+                try
-                stopwatch.Start();
-                using (IDeepSpeech sttClient = new DeepSpeech(model ?? "output_graph.pbmm",
-                    BEAM_WIDTH))
                 {
+                    Console.WriteLine("Loading model...");
+                    stopwatch.Start();
+                    sttClient.CreateModel(
+                        model ?? "output_graph.pbmm",
+                        BEAM_WIDTH);
                     stopwatch.Stop();
 
                     Console.WriteLine($"Model loaded - {stopwatch.Elapsed.Milliseconds} ms");
@@ -85,14 +88,12 @@ namespace CSharpExamples
                         string speechResult;
                         if (extended)
                         {
-                            Metadata metaResult = sttClient.SpeechToTextWithMetadata(waveBuffer.ShortBuffer,
+                            Metadata metaResult = sttClient.SpeechToTextWithMetadata(waveBuffer.ShortBuffer, Convert.ToUInt32(waveBuffer.MaxSize / 2));
-                                Convert.ToUInt32(waveBuffer.MaxSize / 2));
                             speechResult = MetadataToString(metaResult);
                         }
                         else
                         {
-                            speechResult = sttClient.SpeechToText(waveBuffer.ShortBuffer,
+                            speechResult = sttClient.SpeechToText(waveBuffer.ShortBuffer, Convert.ToUInt32(waveBuffer.MaxSize / 2));
-                                Convert.ToUInt32(waveBuffer.MaxSize / 2));
                         }
 
                         stopwatch.Stop();
@@ -103,10 +104,10 @@ namespace CSharpExamples
                     }
                     waveBuffer.Clear();
                 }
-            }
+                catch (Exception ex)
-            catch (Exception ex)
+                {
-            {
+                    Console.WriteLine(ex.Message);
-                Console.WriteLine(ex.Message);
+                }
             }
         }
     }