# YoutubeToText

> Transcripts, subtitles and translations from a YouTube link. The same engine
> is available as a REST API and an MCP server, on the same minutes as the site.

Accuracy is measured monthly and published, rather than claimed: https://youtubetotext.ai/accuracy/

## Developer documentation

- [Start here](https://youtubetotext.ai/docs/): What the API does, how a job works, and where to get a token.
- [Quickstart](https://youtubetotext.ai/docs/quickstart/): Token, one request, one poll, your first transcript.
- [Authentication](https://youtubetotext.ai/docs/authentication/): Bearer tokens, where they come from, and what a 401 means.
- [API reference](https://youtubetotext.ai/docs/api-reference/): Every endpoint, its parameters and what comes back.
- [Jobs and polling](https://youtubetotext.ai/docs/jobs-and-polling/): The twelve states, how often to poll, and reading the result.
- [Errors](https://youtubetotext.ai/docs/errors/): Status codes, the validation shape, and jobs that fail after 200.
- [Minutes and limits](https://youtubetotext.ai/docs/pricing-and-limits/): What a call costs in minutes, and every ceiling worth knowing.
- [Python](https://youtubetotext.ai/docs/python/): A working script, and how this differs from reading captions.
- [n8n, Make, Zapier](https://youtubetotext.ai/docs/integrations/): The same two calls, as nodes, with the wait in between.

## API

Base URL: https://api.youtubetotext.ai
Machine-readable spec: https://youtubetotext.ai/openapi.json
Authentication: a bearer token from https://youtubetotext.ai/account/api/

- `POST /v1/api/transcribe`: Queue one video and get a job id back.
- `GET /v1/api/transcription/{id}`: Read a job: its state while it runs, its text when it is done.
- `POST /v1/api/subtitles`: Queue one video and get SRT and WebVTT cues back.
- `POST /v1/api/burn-subtitles`: Queue one video and get a video with the captions rendered into it.

A queue call answers 200 with a job id before any work happens. Poll the job
until its state is done or failed; a job that failed puts the reason on
`error` and is not charged for. The states are: waiting, preparing, downloading, converting, uploading, processing, merging, aligning, translating, burning, done, failed.

## MCP

Server: https://api.youtubetotext.ai/mcp
Works with Claude, ChatGPT, Cursor and Cline. It signs in through the account,
so no token is needed: https://youtubetotext.ai/developers/mcp/

## Product

- [Transcription](https://youtubetotext.ai/transcription/): transcripts with speaker labels and timestamps
- [Subtitles](https://youtubetotext.ai/subtitles/): SRT and WebVTT, or captions burned into the video
- [Pricing](https://youtubetotext.ai/pricing/): plans, credit packs, and what a minute costs
- [FAQ](https://youtubetotext.ai/faq/)