Clipsmith Audio Transcript

Convert a local audio file to text by running an explicit local transcription command. Use after extracting audio from video.

OctopusGarage Updated

File contents

Clipsmith Audio Transcript

Transcribe a local audio file by invoking a caller-provided local command. This skill does not extract audio from video, create snapshots, run OCR, or write capture bundles.

Execution

cd <clipsmith-repo>/skills/clipsmith-audio-transcript
npx tsx scripts/run.ts \
  --audio_path "/path/to/audio.m4a" \
  --output_dir "$HOME/Downloads/clipsmith-audio-transcript" \
  --transcript_cmd "whisper {audio} --output_format txt --output_dir {output_dir}"

The command template may use:

  • {audio}: absolute audio path
  • {output_dir}: output directory
  • {output_text}: expected text output path

If the command writes stdout but does not create {output_text}, stdout is saved to transcript.txt.

Constraints

  • Accept local audio paths only.
  • Require an explicit transcription command.
  • Do not call network APIs or hardcode a provider SDK.
  • Do not modify the source audio file.

Success Criteria

  1. Source audio exists.
  2. Transcription command exits with status 0.
  3. transcript.txt and transcript.md are written.

OctopusGarage/clipsmith commit 8b19dfc15a

Frequently asked questions

npx get-skill add octopusgarage/clipsmith-audio-transcript