faster-whisper
faster-whisper copied to clipboard
The timeline is advanced by 0.5s~1s
version:1.0.1 system: win10 py310 cuda12 config: whisper-ctranslate2 --model medium --output_format srt --local_files_only True --word_timestamps True --language ko --vad_filter false
Almost every sentence will be displayed 0.5s~1s earlier
Where is the audio sample to reproduce the issue?