Transcribe audio and video locally with word timings (/transcribe)
Transcribe audio or video to text locally — on the Apple GPU with mlx-whisper large-v3, or faster-whisper on CPU/CUDA — with the language decided per stretch of speech, .srt subtitles, word…
/transcribe {file-or-folder} Transcribe locally with word timings. Keep a glossary of names so the model stops turning them into celebrities: {Name1, Name2}. Output .srt beside each file, plus one day.txt with every sentence numbered by clip and order, which I will read as a script.
In Claude Code, paste the prompt as a message, or run /transcribe which does this end to end.
Install
npx skills add fortunto2/solo-factory/transcribeSteps
- Install the skill into Claude Code with the line above (or all of solo-factory: npx skills add fortunto2/solo-factory --all).
- Run it when: You need the speech of a recording as text with timestamps, on your machine.
- The skill asks for what it needs and writes its output into the project; read the summary it prints before trusting the files.
Example
Any ffmpeg-decodable file becomes <name>.txt ([MM:SS] (lang) text), <name>.srt, and on request <name>.words.json with per-word timings, the input a video edit needs. Audio never leaves the machine; the only network step is a one-time model download.
Tips
- Source and the full instructions: github.com/fortunto2/solo-factory/tree/main/skills/transcribe (v1.0.1).
- Skills are prompts, not programs: read SKILL.md once, then you know exactly what the agent will do.
| works with | Claude Code: Runs as a skill or a prompt in the terminal; can read your files and write the result. |
| skill | npx skills add fortunto2/solo-factory/transcribe · source |
| cost | Free. No API, no key. |
| status | Unverified: written from the technique, not yet run end to end by us. |
| source | solo-factory/skills/transcribe (MIT) |
| license | CC BY 4.0 for the text here; skills are MIT. |
| updated | 2026-11-01 |
| for agents | skill-transcribe.md |
Related
Edit real footage into a film with the agent as the editor (/video-edit)
Edit real footage into a film with the agent as the editor — read the day's speech as text, write the cut as a script, lay cutaways that match the words, score it, strip other people's music, and…
Turn an article or PRD into a social media pack (/content-gen)
Use when "create content", "write LinkedIn post", "social media pack", "marketing copy", "promotion content", "posts from this article", "repurpose this post", "посты из статьи", or need soc
Plan short-form growth for an app: founder-led first, creators after a signal (/creator-growth)
Plan growth for a consumer app through short vertical video — founder-led content first, paid UGC creators only after a measured signal.
Strip AI patterns from a script or caption so it reads like a person (/humanize)
Use when "humanize this", "make it sound human", "strip AI patterns", "clean up the copy", or text reads like AI-generated output with em dashes and stock phrases.
Write a landing page with copy, SEO, JSON-LD and a cookieless counter (/landing-gen)
Turn a PRD into a landing page that converts and ships honest. One batch of intake questions, section-by-section copy with A/B headlines, a type, spacing, radius and motion system on Tailwind…
Separate vocals, music and drums from a recording (/stem-split)
Separate a song or a film's soundtrack into stems (vocals, drums, bass, other) with Demucs, locally, on the Apple GPU when there is one.