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

mono-channel error, not just an assertion

Browse Source 
X-DeepSpeech: NOBUILD
This commit is contained in:
josh meyer 2020-10-02 13:25:44 -07:00
parent dd4122a04a
commit afee570f3c
1 changed files with 4 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
training/deepspeech_training/util/audio.py
View File

@ -377,7 +377,8 @@ def get_dtype(audio_format):
def pcm_to_np(pcm_data, audio_format=DEFAULT_FORMAT): def pcm_to_np(pcm_data, audio_format=DEFAULT_FORMAT):
assert audio_format.channels == 1 # only mono supported for now if audio_format.channels != 1:
raise ValueError('Mono-channel audio required')
dtype = get_dtype(audio_format) dtype = get_dtype(audio_format)
samples = np.frombuffer(pcm_data, dtype=dtype) samples = np.frombuffer(pcm_data, dtype=dtype)
samples = samples.astype(np.float32) / np.iinfo(dtype).max samples = samples.astype(np.float32) / np.iinfo(dtype).max
@ -385,7 +386,8 @@ def pcm_to_np(pcm_data, audio_format=DEFAULT_FORMAT):
def np_to_pcm(np_data, audio_format=DEFAULT_FORMAT): def np_to_pcm(np_data, audio_format=DEFAULT_FORMAT):
assert audio_format.channels == 1 # only mono supported for now if audio_format.channels != 1:
raise ValueError('Mono-channel audio required')
dtype = get_dtype(audio_format) dtype = get_dtype(audio_format)
np_data = np_data.squeeze() np_data = np_data.squeeze()
np_data = np_data * np.iinfo(dtype).max np_data = np_data * np.iinfo(dtype).max