# X (Twitter) Video Transcript API

X

Returns the timed transcript of a public X (formerly Twitter) video post from its link or ID, from X's own captions or, when it has none, the video's audio. Each segment has its start and end time, text holds the whole transcript, and language says what was spoken. Reposts, quotes of a video, videos in a post's card and posts with several videos work too. [Overview and pricing](https://socialtap.dev/apis/x/transcript/index.md)

GET `/v1/x/transcript`

## Example

cURL:

```shell
curl -G 'https://socialtap.dev/v1/x/transcript' \
  --data-urlencode 'url=https://x.com/SawyerMerritt/status/2117697749009207386/video/1' \
  -H "x-api-key: $SOCIALTAP_API_KEY"
```

TypeScript SDK:

```typescript
import { SocialTap } from "socialtap";

const client = new SocialTap(); // apiKey defaults to process.env.SOCIALTAP_API_KEY

const res = await client.x.transcript.get({ url: "https://x.com/SawyerMerritt/status/2117697749009207386/video/1" });
console.log(res.data);
console.log(res.credits.charged, res.meta.request_id);
```

Node.js (fetch):

```javascript
const params = new URLSearchParams({
  url: "https://x.com/SawyerMerritt/status/2117697749009207386/video/1",
});
const res = await fetch(`https://socialtap.dev/v1/x/transcript?${params}`, {
  headers: { "x-api-key": process.env.SOCIALTAP_API_KEY ?? "" },
});
const body = await res.json();
if (!res.ok) throw new Error(`${body.error.code}: ${body.error.message}`);
const { data, meta } = body;
console.log(data, meta.credits_charged);
```

Python (requests):

```python
import os

import requests

res = requests.get(
    "https://socialtap.dev/v1/x/transcript",
    params={"url": "https://x.com/SawyerMerritt/status/2117697749009207386/video/1"},
    headers={"x-api-key": os.environ["SOCIALTAP_API_KEY"]},
)
body = res.json()
if not res.ok:
    raise RuntimeError(f"{body['error']['code']}: {body['error']['message']}")
data, meta = body["data"], body["meta"]
print(data, meta["credits_charged"])
```

Go (net/http):

```go
package main

import (
	"encoding/json"
	"fmt"
	"log"
	"net/http"
	"net/url"
	"os"
)

func main() {
	q := url.Values{}
	q.Set("url", "https://x.com/SawyerMerritt/status/2117697749009207386/video/1")
	req, err := http.NewRequest("GET", "https://socialtap.dev/v1/x/transcript?"+q.Encode(), nil)
	if err != nil {
		log.Fatal(err)
	}
	req.Header.Set("x-api-key", os.Getenv("SOCIALTAP_API_KEY"))
	res, err := http.DefaultClient.Do(req)
	if err != nil {
		log.Fatal(err)
	}
	defer res.Body.Close()
	var body struct {
		Data  json.RawMessage `json:"data"`
		Meta  map[string]any  `json:"meta"`
		Error *struct {
			Code    string `json:"code"`
			Message string `json:"message"`
		} `json:"error"`
	}
	if err := json.NewDecoder(res.Body).Decode(&body); err != nil {
		log.Fatal(err)
	}
	if body.Error != nil {
		log.Fatalf("%s: %s", body.Error.Code, body.Error.Message)
	}
	fmt.Println(string(body.Data), body.Meta["credits_charged"])
}
```

200 OK:

```json
{
  "data": {
    "available_languages": [
      "en"
    ],
    "duration_ms": 95266,
    "id": "2117697749009207386",
    "is_auto_generated": true,
    "language": "en",
    "platform": "x",
    "platform_specific": {
      "x": {
        "caption_url": "https://video.twimg.com/subtitles/amplify_video/2117697012804071425/0/xVEM4xrAnRm8mGa7.vtt",
        "media_index": 0,
        "video_post_id": "2117697749009207386"
      }
    },
    "segments": [
      {
        "end_ms": 1080,
        "start_ms": 80,
        "text": "Hi"
      },
      {
        "end_ms": 2561,
        "start_ms": 340,
        "text": "I'm a virtual assistant calling for Sawyer Merritt"
      },
      {
        "end_ms": 4221,
        "start_ms": 2641,
        "text": "on a recorded line to ask about a"
      }
    ],
    "source": "captions",
    "text": "Hi I'm a virtual assistant calling for Sawyer Merritt on a recorded line to ask about a door handle return Is Barbara available? Speaking Hi Barbara Sawyer wanted me to ask if you could figure out which shipping carrier would be cheapest for sending back that door handle you two …",
    "url": "https://x.com/SawyerMerritt/status/2117697749009207386"
  },
  "meta": {
    "request_id": "req_3f9c2a7d1e8b4c6a9f0e2d4b6a8c0e1f",
    "platform": "x",
    "source": "captions",
    "is_cached": false,
    "credits_charged": 1,
    "latency_ms": 1000,
    "fetched_at": "2026-10-25T12:00:00Z",
    "input": {
      "kind": "post",
      "id": "2117697749009207386",
      "url": "https://x.com/SawyerMerritt/status/2117697749009207386"
    },
    "credits_remaining": 9999,
    "client_request_id": null,
    "warnings": []
  }
}
```

Long arrays and strings are shortened for display; real responses are complete.

> [!NOTE]
>
> **Good to know**
>
> - X captions many recent videos itself, in the spoken language (meta.source = captions, is_auto_generated true). Videos X has no captions for are transcribed by speech-to-text (meta.source = asr, priced per minute) up to a length limit.
> - lang picks X's caption track in that language; X has one per video, in its spoken language, so another lang returns that track and language says which.
> - A link ending in /video/2 transcribes the post's second attachment; without it, the first video. platform_specific.x.video_post_id and media_index say which video was transcribed.
> - A video X shows in the post's card (an ad's or a publisher's) is the post's video. GIFs have no sound and return no_transcript, as do live broadcasts. Text and photo posts return invalid_input.
> - Deleted posts and posts from suspended accounts return not_found. Posts from protected accounts and age-restricted posts return private_or_unavailable.

## Authorizations

`x-api-key`stringheaderrequired

Your API key. [Authentication](https://socialtap.dev/docs/authentication/index.md)

## Query parameters

| Name | Type | Description | Example |
| --- | --- | --- | --- |
| `url` required | `string`<br>max 2,048 chars | X video post: a link (x.com or twitter.com, /\<username>/status/\<id> with or without /video/\<n>, which picks the post's nth attachment, /i/web/status/\<id>, /i/status/\<id>, old /statuses/ and #! links, mobile.twitter.com, fxtwitter.com and vxtwitter.com) or a bare numeric post ID. A repost transcribes the reposted video, and a quote post without a video of its own the video of the post it quotes. | `https://x.com/TheoVon/status/1916982720317821050` |
| `lang` optional | `string`<br>bcp47 | Preferred language: a BCP-47 tag such as en or pt-BR (case-insensitive; _ is read as -). Picks X's caption track in that language when the video has one. X captions a video in its spoken language only, so for another language you get that track, and language says which. For a video X has no captions for, it is the language speech-to-text listens for (detected without it). | `en` |

`url` takes X post links; another kind of link, or another platform's, fails free with [`invalid_input`](https://socialtap.dev/docs/errors/index.md#invalid_input). Unknown parameters are rejected with a suggestion, never ignored ([Requests](https://socialtap.dev/docs/requests/index.md)).

**Parameters every endpoint accepts**

| Name | Type | Description | Example |
| --- | --- | --- | --- |
| `max_age` optional | `integer`<br>min 0 | Accept a cached response up to this many seconds old. Cache hits cost 0 credits. Omit to always fetch fresh. The Cache-Control: max-age=N request header is equivalent; max_age wins when both are sent. | — |
| `fields` optional | `string` | Return only these fields of each item: comma-separated paths in the schema's notation (id,text,metrics.view_count,segments\[\].text; a parent path keeps the whole object). Unknown paths are rejected. Applied after the cache, so it changes neither the price nor cache hits; page and meta are always returned whole. | — |

## Response

### 200

`{data, meta}`. `data` is a [transcript](https://socialtap.dev/resources/transcript/index.md) object; `meta` describes the request ([meta](https://socialtap.dev/docs/reference/index.md#meta)). Fields only X has are under `platform_specific.x`.

| Field | Type | Description |
| --- | --- | --- |
| `data.platform` | `string` | Platform the media is hosted on. |
| `data.id` | `string` | Platform-native ID of the post (equals post.id); of an ad library entry, its library ID (equals ad.id). |
| `data.url` | `string` (uri) | Canonical URL of the post (of an ad library entry, its library page). |
| `data.language` | `string` or `null` | BCP-47 language of the returned transcript; null when unknown. |
| `data.source` | `string` | captions = platform caption track; asr = speech-to-text on the audio. New values may be added.<br>One of `captions`, `asr` |
| `data.is_auto_generated` | `boolean` | True when the track was machine-generated (platform auto-captions or ASR). |
| `data.available_languages` | `string[]` | BCP-47 caption languages the platform offers for this post; \[\] if none or not listed. |
| `data.duration_ms` | `integer` or `null` | Media duration in milliseconds; null when unknown. |
| `data.segments` | `object[]` | Timed cues in order; never empty on success. |
| `data.segments[].start_ms` | `integer` | Start time in milliseconds from the beginning of the media. |
| `data.segments[].end_ms` | `integer` | End time in milliseconds from the beginning of the media. |
| `data.segments[].text` | `string` | Cue text, whitespace-normalized. |
| `data.text` | `string` | Full transcript: segment texts joined with single spaces. |
| `data.platform_specific` | `object` | Fields only this platform has, under its ID (platform_specific.\<platform>); each null when unknown. |
| `data.platform_specific.x` | `object` or `null` | An X transcript's own fields. |
| `data.platform_specific.x.video_post_id` | `string` or `null` | ID of the post the transcribed video is attached to: the post itself, the reposted post for a repost, or the quoted post for a quote without a video of its own; null when unknown. |
| `data.platform_specific.x.media_index` | `integer` or `null` | Index of the transcribed video in the media of the post video_post_id names (0 for the first attachment, as x.post's media); null when unknown. |
| `data.platform_specific.x.caption_url` | `string` (uri) or `null` | The WebVTT file of X's caption track used (public, unsigned); null for speech-to-text. |

not_found, private_or_unavailable and no_transcript are item-level: the request was fine, the item can't be returned.

### 400

400 Bad Request: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `invalid_input` | 400 | No | A parameter is missing or malformed, or the link is not one this endpoint takes. | Fix the parameter named in param as the message says. For a link of the wrong kind, call an endpoint listed in details.endpoints, when there is one. |

### 401

401 Unauthorized: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `unauthorized` | 401 | No | The x-api-key header is missing, or its key is unknown or revoked. On the dashboard's routes: the session cookie is missing or expired. | Send a valid API key in the x-api-key header. No key yet? A person gets one at details.signup_url; an AI agent follows details.agent_auth_url. |

### 402

402 Payment Required: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `insufficient_credits` | 402 | No | The account does not have enough credits for this request. | Have a person buy credits at details.purchase_url (or make them a checkout link with details.checkout), then repeat the request. Do not retry automatically. |
| `spend_limit_reached` | 402 | No | The API key reached its spend limit for this period. The account may still have credits. | Wait until details.resets_at, or raise the key's spend limit (PATCH /v1/api_keys/{id}). Do not retry automatically. |

### 403

403 Forbidden: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `forbidden` | 403 | No | The API key is valid but may not make this request, and details.reason says why: the account is suspended, the key lacks the route's scope, has expired, or was used from an IP outside its allowlist, or a rotated key's previous secret was used for a call that needs the admin scope. A running job stops with it too when its account is suspended or the key that created it is revoked or expires. It is about the key, never the item: a private item is private_or_unavailable, also 403, so branch on the code. | Act on details.reason. suspended: contact [support](mailto:support@socialtap.dev). scope: use a key with details.required_scope. previous_secret: use the new secret. key_expired or key_revoked: create a new key. ip_not_allowed: call from an allowed IP. |
| `private_or_unavailable` | 403 | No | The item exists but is not public: it is private, age- or region-restricted, or needs a login to view. It shares 403 with forbidden (which is about the key), so branch on the code. | Only public items can be fetched. Do not retry. |

### 404

404 Not Found: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `not_found` | 404 | No | The item does not exist or was deleted. On GET /v1/requests/{id}/response: the response was not kept (details.reason says why). | Check the link or ID. Do not retry. |
| `no_transcript` | 404 | No | The video or audio exists but has no transcript: no caption track and no speech to transcribe. | Do not retry. If the platform offers other languages, try one with lang. |

### 429

429 Too Many Requests: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `rate_limited` | 429 | Yes, later | The account passed its runaway-protection ceiling: a high per-account limit that stops a broken client loop (plans have no rate limit). Requests without an API key (signup, the MCP server's tool listing and find_endpoint) are limited per IP address instead. It also stops an account whose failed requests in the last hour cost far more than its successful ones paid for: failures are free within a limit (the message says so), and successes raise it. | Wait the seconds in the retry-after header, then retry. If your steady traffic needs more, ask [support](mailto:support@socialtap.dev) to raise the ceiling; without an API key, send one. |

### 451

451: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `unavailable_for_legal_reasons` | 451 | No | We removed this item after a copyright notice or another legal request, so we don't return it. The request was free. A list leaves such items out instead, with an items_omitted warning. | Do not retry. The item stays unavailable unless the notice is withdrawn. |

### 500

500 Internal Error: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `internal` | 500 | Yes, later | An unexpected error on SocialTap's side. | Retry with backoff. If it keeps happening, contact [support](mailto:support@socialtap.dev) with the request_id. |

### 502

502 Bad Gateway: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `upstream_error` | 502 | Yes, later | The platform failed, or answered with something we could not read. | Retry with backoff. Failed requests are free. |

### 503

503 Unavailable: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `upstream_blocked` | 503 | Yes, later | The platform refused every attempt, our own retries included. | Retry later with backoff. Failed requests are free. |
| `overloaded` | 503 | Yes, later | SocialTap was at capacity, or paused by its operators, and did not start the request. Nothing was fetched or charged. | Retry after the seconds in the retry-after header. Failed requests are free. |

## Pricing

From **1 credit** per request, up to 122 credits. Charged only for successful requests. Errors are free. The price depends on the source that produced the result, reported as `meta.source`; sources are tried cheapest first: Captions → Speech-to-text.

| Source | Price per successful request | Per 1,000 requests |
| --- | --- | --- |
| Captions `captions` | 1 credit | $0.75–$1.97 |
| Speech-to-text `asr` | 2 credits + 2 per started minute of media<br>up to 122 credits (60 started minutes of media) | $1.50–$3.93 + $1.50–$3.93 per started minute of media |

Example: a result from Speech-to-text `asr` with 3 started minutes of media costs 2 + 2 × 3 = **8 credits**.

**Balance, sources and what you were charged**

Your balance must be at least **1 credit** to start a request; with less, it is refused with [`insufficient_credits`](https://socialtap.dev/docs/errors/index.md#insufficient_credits) (HTTP 402) and costs nothing. Only that minimum is set aside when the request starts. One request costs at most **122 credits**, but that amount is never locked up in advance: right before costlier work, SocialTap sets aside more of your balance, the exact price when the number of started minutes of media is known beforehand, otherwise as much as your balance allows (and the work is limited to what that pays for). It then charges the actual price and releases the rest. If the result would cost more than your balance covers, the request is refused with [`insufficient_credits`](https://socialtap.dev/docs/errors/index.md#insufficient_credits) as well, and costs nothing.

| meta.source | Source | What it means |
| --- | --- | --- |
| `captions` | Captions | The platform's own caption track (uploaded or auto-generated). |
| `asr` | Speech-to-text | Transcribed by SocialTap from the media's audio when the platform has no captions. |

`meta.credits_charged` and the `x-credits-charged` header show what you paid. [How credits work](https://socialtap.dev/docs/credits/index.md) · [Packs](https://socialtap.dev/pricing/index.md)

Endpoint ID `x.transcript` · operationId `x.transcript.get` · [Product overview](https://socialtap.dev/apis/x/transcript/index.md)

---

[HTML version](https://socialtap.dev/docs/reference/x/transcript/) · [Docs index for AI agents](https://socialtap.dev/llms.txt) · [All docs in one file](https://socialtap.dev/llms-full.txt)
