EXTENSIONS

Transcribe meeting audio locally with Whisper (faster-whisper, large-v3 by default) in a container under Apple's container CLI or Docker. Writes a timestamped text transcript plus SRT, WebVTT and JSON with word-level timestamps; the image builds itself on first use and model weights are cached between runs.

Apply pyannote/speaker-diarization-3.1 to a Whisper transcript JSON and tag each segment with a speaker label. Requires Python + pyannote.audio and a Hugging Face token (gated model). Falls back gracefully — a failed diarization never blocks the pipeline; the original transcript is returned unchanged.

Transcribe audio files using mlx_whisper on Apple Silicon. Runs the mlx_whisper CLI, captures the JSON output, and stores the full transcript (segments + language) as a swamp resource for downstream pipeline steps. Requires mlx_whisper installed (pip install mlx-whisper).

Chunk a Whisper transcript and summarize it into a Granola-style meeting note via a local Ollama LLM. Handles long recordings with greedy segment-boundary chunking and an LLM merge pass. Pure TypeScript — no Python, no subprocess. Also provides combine_notes (merge handwritten + analysis) and write_note (write markdown to disk) for full pipeline workflows.