> Parel documentation. Index of every page: https://parel.si/docs/llms.txt. This page in HTML: https://parel.si/docs/tools/get_youtube_transcript

# get_youtube_transcript

> The spoken words of a YouTube video as text, for summarising or quoting. ₹0.28 per video; no transcript is ₹0.

## What it does

The spoken words of one YouTube video: `data.transcript_only_text`, with `videoId`, `url` and `language`. Use it to summarise a video or find what was said. There are no timestamps.

For the title, views and channel use [`get_youtube`](https://parel.si/docs/tools/get_youtube.md), for comments [`get_youtube_comments`](https://parel.si/docs/tools/get_youtube_comments.md), and to find a video [`search_youtube`](https://parel.si/docs/tools/search_youtube.md).

India: Hindi videos come back in Devanagari when the video has captions. In the live check, three Shark Tank India videos did.

## Inputs

Generated from the contract captured from staging on 2026-11-05.

| Field | Type | Required | Description |
|---|---|---|---|
| `url` | string | yes | The video's YouTube URL, long or short. |
| `language` | string | no | Two-letter language code ("en", "hi"). If the video has no transcript in that language, the call finds nothing and is free. Leave it out for the video's own language. |

## What drives the cost

Nothing you can turn: flat per video whatever its length. A video with no captions, a language it has no transcript in, or a video that does not exist is `class: "ok_free"`, ₹0.

Long videos are the catch. Over MCP, text past about 48,000 characters, roughly forty minutes of speech, is cut with a `_truncated` note. The REST API returns it all: a three-and-a-half-hour video in the live check was 257,000 characters and took 21 seconds.

## Price today

| | |
|---|---|
| Price today | **₹0.28** per video (`cost_paise: 28`) at ₹93.93 per US dollar on 17 November 2026 |
| Hold before the call | ₹0.42 (`hold_paise: 42`); when the call settles the real price is charged and the unused part is released |
| ₹0 when | it fails, or the provider answers but there is nothing in it (an empty list, no email, no page): `class: "ok_free"`, ₹0 |
| Timeout | 60 s, then `upstream_timeout` (₹0, retryable) |
| Over MCP | `run_tool` with `tool: "get_youtube_transcript"` |
| Over REST | `POST https://api.parel.si/v1/run` with `"tool": "get_youtube_transcript"` |
| Confirmation | not needed (the estimate is under the ₹200 threshold) |

## Request

```bash
curl https://api.parel.si/v1/run \
  -H "Authorization: Bearer $PAREL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tool":"get_youtube_transcript","input":{"url":"https://www.youtube.com/watch?v=bjVIDXPP7Uk"}}'
```

## What comes back

`data`. A real answer recorded during the live check of 2026-11-05, with the text shortened:

```json
{
  "videoId": "bjVIDXPP7Uk",
  "url": "https://www.youtube.com/watch?v=bjVIDXPP7Uk",
  "language": "English",
  "transcript_only_text": " Welcome  back  to  the  hell  farm  and  the \n backyard  trails.  We  built  these  jumps \n two  years  ago  and  last  year  we  just  kind \n of  rebuilt  them  and  this  year  I  promised \n to  add  to  them  which  we're  going  to  do. \n So  we've  only …"
}
```

## Notes

- Flat price per video; nothing found is free.
- No timestamps: the answer is the words as one text.
- Over MCP, text past about 48,000 characters (roughly forty minutes of speech) is cut with a _truncated note; the REST API returns it all.
- A long video can take twenty seconds or more.

## Siblings

- [`get_youtube`](https://parel.si/docs/tools/get_youtube.md) for a video's title, views and channel.
- [`get_youtube_comments`](https://parel.si/docs/tools/get_youtube_comments.md) for what viewers said.
- [`search_podcasts`](https://parel.si/docs/tools/search_podcasts.md) for podcast episodes by what was said.
