transcribe-audio

Transcribe video audio into timestamped JSON transcripts with word-level timing using WhisperX, which processes full video files directly without separate audio extraction and preserves original timeline alignment for editing and analysis.

Yes, this Skill supports parallel transcription of multiple files, allowing you to process video batches efficiently and output all transcripts to a shared library directory with preserved timing.

Transcripts are generated as JSON with word-level timing data, including video metadata, formatted for downstream visual analysis, captioning, rough cut preparation, and automated processing pipelines.

WhisperX processes full video files directly without requiring separate audio extraction, preserving the original video timeline including leading silence and maintaining accuracy for editing workflows.

The Skill reads the language code from your library.yaml configuration file, automatically applying the specified language setting to transcription without manual parameter configuration per file.

WhisperX accuracy depends on audio quality and language clarity; very long videos may require extended processing time, though parallel processing helps mitigate batch transcription delays.