speech-to-text

To transcribe local video files to VTT subtitles offline, run the script with a video file or directory path. It uses local Whisper to generate timestamped VTT subtitle and TXT transcript files saved in the same directory.

For Whisper transcription on a Mac without a GPU, the best way is using the medium model. The large model is too slow for non-GPU hardware, while GPU-enabled machines can opt for the large model.

Whisper handles existing VTT subtitle files by skipping the video transcription process entirely. This idempotent behavior prevents redundant processing and preserves previously generated timestamped subtitles.

Yes, the Whisper speech-to-text tool supports MKV and WebM video formats. It processes local video content from these files and outputs readable transcripts alongside timestamped VTT subtitles.

Yes, you can batch process a directory of videos for subtitle generation. Pass a directory path to the script, and it applies Whisper transcription to all contained videos, skipping any with existing VTT files.

Local video transcription skips files that already have a VTT subtitle file in the same directory. This idempotent check avoids duplicating work and ensures only videos without existing text transcripts are processed.