AI Video Transcriber skill
The skill turns a video or podcast link or file into a transcript and summary using platform subtitles and Whisper.
- 3,411 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Transcript text goes to the chosen model API, and the script downloads media from links and runs local programs.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/ai-video-transcriberDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Transcript text goes to the chosen model API, and the script downloads media from links and runs local programs.
This is third-party code. Review the repository files before installing.
What it does
The AI Video Transcriber repository ships a `video-transcribe` skill and a Codex plugin. The skill runs `transcribe.py` with no server or browser: it takes platform subtitles (YouTube and 30+ other services) or, when absent, transcribes audio locally with Faster-Whisper. It then cleans up the text, translates it and writes a summary in one of 11 languages. Input can be a link, audio, video or text files. The agent receives JSON with paths to the transcript, summary and translation files. Model steps need a key for any OpenAI-compatible API; `--no-llm` works without one.
Who it is for. Researchers, editors and developers who ask an agent to digest a long video or podcast and return text with a summary.
Good fit when
- A lecture from YouTube or Bilibili needs a transcript.
- An Apple Podcasts episode should become a summary in Russian.
- A local `.mp4` meeting recording must become text.
- The source video is kept alongside the transcript.
Not a fit when
- The video is protected and the platform exposes no subtitles or stream.
- Data must not go to an external model API.
- You need real-time transcription.
Example request
Transcribe this YouTube video and write a summary in Russian without keeping the source file.Limitations
The skill is run from the repository root and needs a Python virtualenv, `ffmpeg` and the dependencies from `requirements.txt`. Optimization, translation and summary need an OpenAI-compatible API key, and the text goes to that provider. Whisper on CPU is slow, and downloads depend on platform rules. The last commit is dated 11 October 2026. Downloads from some platforms may fail from a region without access to them.
How to disable. Remove the skill folder and the Codex plugin. Delete the `venv` environment and temporary files in `./temp` separately and unset the API key from your environment.
FAQ
Is an API key needed for transcription?
No, a raw transcript needs no key: run with `--no-llm`. An OpenAI-compatible key is required for text cleanup, translation and the summary.
How do I choose the summary language?
Pass `-l` or `--summary-language` with a code: `en`, `zh`, `es`, `fr`, `de`, `it`, `pt`, `ru`, `ja`, `ko` or `ar`. The default is `en`.
What if I only want text without the video?
Add `--no-video` and the source clip is not kept. `--whisper-model small` or larger raises Whisper accuracy; the default is `base`.
What exit codes does the command return?
`0` means success, `2` bad input, `1` a download or transcode failure.
Which inputs are accepted?
A platform link, files in `.mp3`, `.mp4`, `.m4a`, `.wav`, `.webm`, `.mkv`, `.ogg`, `.flac` or a `.txt` text. FFmpeg normalizes the media.
Discussion
No questions or reviews yet.
Related
SkillsA Codex and Claude Code skill: subtitles and a summary from a Douyin, Xiaohongshu, Bilibili or YouTube link or a local file
YouTube Skills
YouTube Skills for AI Agents
A set of skills that gives an agent YouTube transcripts, video and channel search, and playlist extraction over one API
Whisper is a CLI for audio transcription, with model sizes for different hardware and output in the original spoken language.
Skill turning audio, video and YouTube links into transcripts and subtitles via your own local Whisper Txtify in Docker.
Twelve skills teach an agent to read YouTube through the TranscriptOut service: transcripts, search and channels in one HTTPS call
MCP server that gives an agent a YouTube video's transcript from its link, with language choice, timestamps and ads filtered out
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| AI Video Transcriber (this card) | no | 3,411 | ||
| YouTube Transcript Server | no | not required | 603 | serves YouTube subtitles over MCP; this adds a Whisper fallback, translation and a summary |
| whisper.cpp | no | 55,136 | a local speech recognition engine; the skill builds a link-to-summary pipeline around it |
README in short
Native platform subtitles come first, and Whisper runs only when they are missing. The source video downloads in parallel with transcription.