From e9a082dcf27647eb52585cda6f115454c0ac6856 Mon Sep 17 00:00:00 2001 From: Guillaume Klein Date: Thu, 6 Apr 2023 11:54:40 +0200 Subject: [PATCH] Keep segment timestamps aligned with words timestamps after VAD (#119) --- faster_whisper/transcribe.py | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/faster_whisper/transcribe.py b/faster_whisper/transcribe.py index bbff104..b077d8b 100644 --- a/faster_whisper/transcribe.py +++ b/faster_whisper/transcribe.py @@ -776,14 +776,18 @@ def restore_speech_timestamps( end=ts_map.get_original_time(word.end, chunk_index), ) words.append(word) - else: - words = segment.words - segment = segment._replace( - start=ts_map.get_original_time(segment.start), - end=ts_map.get_original_time(segment.end), - words=words, - ) + segment = segment._replace( + start=words[0].start, + end=words[-1].end, + words=words, + ) + + else: + segment = segment._replace( + start=ts_map.get_original_time(segment.start), + end=ts_map.get_original_time(segment.end), + ) yield segment