Reword a comment

This commit is contained in:
Purfview
2025-10-24 10:30:34 +01:00
committed by GitHub
parent 2389039408
commit b15a216b47

View File

@@ -379,7 +379,7 @@ def get_speech_timestamps(audio: torch.Tensor,
prev_end = next_start = temp_end = 0
possible_ends = []
else:
# Standard cut at max_speech_duration_s: prefer last valid silence (prev_end) if available
# Legacy max-speech cut (use_max_poss_sil_at_max_speech=False): prefer last valid silence (prev_end) if available
if prev_end:
current_speech['end'] = prev_end
speeches.append(current_speech)