transcribe

To transcribe session audio files, the Skill uses whisper-ctranslate2 to process .aac files and generate a merged transcript with timestamps and speaker labeling. It reads session notes for the date and outputs per-speaker .tsv files alongside the full text.

Yes, you can generate speaker-labeled transcripts from .aac audio files. The transcription workflow processes the audio directory and outputs individual .tsv files for each speaker, plus a combined transcript file containing both timestamps and speaker assignments.

Before running the transcription workflow, you need a session directory containing .aac audio files under a transcriptions subfolder and a notes.md file providing the session date. The script reads these inputs to automate the whisper-ctranslate2 processing.

The best way to document meetings as searchable text is using automated transcription with speaker labeling. This Skill converts session .aac audio files into structured .tsv files and a merged timestamped transcript, making the session content easily searchable.

Yes, the transcription process supports merging multiple audio files from a single session. It lists all .aac files in the transcriptions subfolder and runs a merge script to combine them into a single timestamped transcript with full speaker labeling.

The transcript generation requires a notes.md file because the workflow reads it to determine the session date. This date metadata is necessary to correctly organize and process the .aac audio files located in the transcriptions subfolder.