deepgeek/silero-vad
2
0
Fork 0
mirror of https://github.com/snakers4/silero-vad.git synced 2026-02-05 18:09:22 +08:00
Code Issues Packages Projects Releases Wiki Activity

add min_silence_duration

Browse Source
This commit is contained in:
adamnsandle
2021-02-05 15:08:35 +00:00
parent db0503c149
commit ab54e692c3
1 changed files with 17 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
utils_vad.py
View File

@@ -60,6 +60,7 @@ def get_speech_ts(wav: torch.Tensor,
batch_size: int = 200, batch_size: int = 200,
num_samples_per_window: int = 4000, num_samples_per_window: int = 4000,
min_speech_samples: int = 10000, #samples min_speech_samples: int = 10000, #samples
min_silence_samples: int = 8000,
run_function=validate, run_function=validate,
visualize_probs=False): visualize_probs=False):
@@ -95,20 +96,31 @@ def get_speech_ts(wav: torch.Tensor,
smoothed_probs = [] smoothed_probs = []
speech_probs = outs[:, 1] # this is very misleading speech_probs = outs[:, 1] # this is very misleading
temp_end = 0
for i, predict in enumerate(speech_probs): # add name for i, predict in enumerate(speech_probs): # add name
buffer.append(predict) buffer.append(predict)
smoothed_prob = (sum(buffer) / len(buffer)) smoothed_prob = (sum(buffer) / len(buffer))
if visualize_probs: if visualize_probs:
smoothed_probs.append(float(smoothed_prob)) smoothed_probs.append(float(smoothed_prob))
if (smoothed_prob >= trig_sum) and temp_end:
temp_end=0
if (smoothed_prob >= trig_sum) and not triggered: if (smoothed_prob >= trig_sum) and not triggered:
triggered = True triggered = True
current_speech['start'] = step * max(0, i-num_steps) current_speech['start'] = step * max(0, i-num_steps)
continue
if (smoothed_prob < neg_trig_sum) and triggered: if (smoothed_prob < neg_trig_sum) and triggered:
current_speech['end'] = step * i if not temp_end:
if (current_speech['end'] - current_speech['start']) > min_speech_samples: temp_end = step * i
speeches.append(current_speech) if step * i - temp_end < min_silence_samples:
current_speech = {} continue
triggered = False else:
current_speech['end'] = temp_end
if (current_speech['end'] - current_speech['start']) > min_speech_samples:
speeches.append(current_speech)
temp_end = 0
current_speech = {}
triggered = False
continue
if current_speech: if current_speech:
current_speech['end'] = len(wav) current_speech['end'] = len(wav)
speeches.append(current_speech) speeches.append(current_speech)