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

Merge pull request #528 from snakers4/adamnsandle

Browse Source 
add neg_threshold parameter explicitly
This commit is contained in:
Dimitrii Voronin
2024-08-22 16:39:33 +03:00
committed by GitHub
parent 36854a90db 4109b107c1
commit 199de226e5
1 changed files with 15 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/silero_vad/utils_vad.py
View File

@@ -195,6 +195,7 @@ def get_speech_timestamps(audio: torch.Tensor,
return_seconds: bool = False, return_seconds: bool = False,
visualize_probs: bool = False, visualize_probs: bool = False,
progress_tracking_callback: Callable[[float], None] = None, progress_tracking_callback: Callable[[float], None] = None,
neg_threshold: float = None,
window_size_samples: int = 512,): window_size_samples: int = 512,):
""" """
@@ -237,6 +238,9 @@ def get_speech_timestamps(audio: torch.Tensor,
progress_tracking_callback: Callable[[float], None] (default - None) progress_tracking_callback: Callable[[float], None] (default - None)
callback function taking progress in percents as an argument callback function taking progress in percents as an argument
neg_threshold: float (default = threshold - 0.15)
Negative threshold (noise or exit threshold). If model's current state is SPEECH, values BELOW this value are considered as NON-SPEECH.
window_size_samples: int (default - 512 samples) window_size_samples: int (default - 512 samples)
!!! DEPRECATED, DOES NOTHING !!! !!! DEPRECATED, DOES NOTHING !!!
@@ -298,6 +302,8 @@ def get_speech_timestamps(audio: torch.Tensor,
triggered = False triggered = False
speeches = [] speeches = []
current_speech = {} current_speech = {}
if neg_threshold is None:
neg_threshold = threshold - 0.15 neg_threshold = threshold - 0.15
temp_end = 0 # to save potential segment end (and tolerate some silence) temp_end = 0 # to save potential segment end (and tolerate some silence)
prev_end = next_start = 0 # to save potential segment limits in case of maximum segment size reached prev_end = next_start = 0 # to save potential segment limits in case of maximum segment size reached
@@ -334,7 +340,7 @@ def get_speech_timestamps(audio: torch.Tensor,
if (speech_prob < neg_threshold) and triggered: if (speech_prob < neg_threshold) and triggered:
if not temp_end: if not temp_end:
temp_end = window_size_samples * i temp_end = window_size_samples * i
if ((window_size_samples * i) - temp_end) > min_silence_samples_at_max_speech : # condition to avoid cutting in very short silence if ((window_size_samples * i) - temp_end) > min_silence_samples_at_max_speech: # condition to avoid cutting in very short silence
prev_end = temp_end prev_end = temp_end
if (window_size_samples * i) - temp_end < min_silence_samples: if (window_size_samples * i) - temp_end < min_silence_samples:
continue continue