EngineX-Iluvatar/enginex-mr_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

Avoid adding tail padding for VAD in generate-subtitles.py (#1674)

Browse Source
This commit is contained in:
Fangjun Kuang
2025-01-03 10:37:39 +08:00
committed by GitHub
parent a00d3b4821
commit a4365dad82
1 changed files with 12 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
python-api-examples/generate-subtitles.py
View File

@@ -516,16 +516,13 @@ def main():
is_eof = False is_eof = False
# TODO(fangjun): Support multithreads # TODO(fangjun): Support multithreads
while True: while not is_eof:
# *2 because int16_t has two bytes # *2 because int16_t has two bytes
data = process.stdout.read(frames_per_read * 2) data = process.stdout.read(frames_per_read * 2)
if not data: if not data:
if is_eof: vad.flush()
break
is_eof = True is_eof = True
# pad 1 second at the end of the file for the VAD else:
data = np.zeros(1 * args.sample_rate, dtype=np.int16)
samples = np.frombuffer(data, dtype=np.int16) samples = np.frombuffer(data, dtype=np.int16)
samples = samples.astype(np.float32) / 32768 samples = samples.astype(np.float32) / 32768
@@ -536,9 +533,6 @@ def main():
vad.accept_waveform(buffer[:window_size]) vad.accept_waveform(buffer[:window_size])
buffer = buffer[window_size:] buffer = buffer[window_size:]
if is_eof:
vad.flush()
streams = [] streams = []
segments = [] segments = []
while not vad.empty(): while not vad.empty():