# Transcripts and on-screen text (/docs/transcripts) Read what is said and what is written on screen, in videos and posts, from one call A video holds words in two places: what people say, and what is written on screen. Each has its own calls. This page lists them, prices one for free and says what happens when a video has no words to read. ## Price a transcript for free ```bash title="cURL" curl -G "https://www.socialcrawl.dev/v1/utility/estimate" \ -H "x-api-key: sc_your_key_here" \ --data-urlencode "id=tiktok/post/transcript" \ --data-urlencode "url=https://www.tiktok.com/@stoolpresidente/video/7499229683859426602" ``` ```json title="Response (trimmed)" { "success": true, "data": { "id": "tiktok/post/transcript", "valid": true, "hold": 10, "pricing": "flat", "formula": "10 credits per call, flat." }, "credits_used": 0 } ``` The estimate costs nothing. A transcript call is priced per video, so a long video costs the same as a short one. ## What is said The transcript endpoints return the spoken words of a video: | Platform | Endpoint | Path | Credits | | --- | --- | --- | --- | | Facebook | [Get a Facebook Ad Library video ad transcript](/platforms/facebook/adlibrary-ad-transcript) | `/v1/facebook/adlibrary/ad/transcript` | 10 | | Facebook | [Get Facebook video transcript](/platforms/facebook/post-transcript) | `/v1/facebook/post/transcript` | 10 | | Instagram | [Get Instagram media transcript](/platforms/instagram/media-transcript) | `/v1/instagram/media/transcript` | 10 | | LinkedIn | [Get a LinkedIn post video transcript](/platforms/linkedin/post-transcript) | `/v1/linkedin/post/transcript` | 10 | | Reddit | [Get a Reddit video post transcript](/platforms/reddit/post-transcript) | `/v1/reddit/post/transcript` | 10 | | Rumble | [Get a Rumble video transcript](/platforms/rumble/video-transcript) | `/v1/rumble/video/transcript` | 10 | | TikTok | [Get TikTok video transcript](/platforms/tiktok/post-transcript) | `/v1/tiktok/post/transcript` | 10 | | Twitter/X | [Get Twitter video transcript](/platforms/twitter/tweet-transcript) | `/v1/twitter/tweet/transcript` | 10 | | YouTube | [Up to 100 YouTube video ids → one transcript per row, failed ids refunded.](/platforms/youtube/transcripts) | `/v1/youtube/transcripts` | 3-300 (metered) | | YouTube | [Get YouTube video transcript](/platforms/youtube/video-transcript) | `/v1/youtube/video/transcript` | 3 | On TikTok, `use_ai_as_fallback=true` transcribes the audio when the video has no caption track, at the same price. [Video transcription](/docs/recipes/video-transcription-api.md) wraps these calls in one function. ## What is written on screen Creators often put text on the video itself, which no caption or transcript carries. These endpoints read it: | Platform | Endpoint | Path | Credits | | --- | --- | --- | --- | | Instagram | [Get Instagram post on-screen text](/platforms/instagram/media-screen-text) | `/v1/instagram/media/screen-text` | 5-10 (metered) | | TikTok | [Get TikTok video on-screen text](/platforms/tiktok/video-screen-text) | `/v1/tiktok/video/screen-text` | 5-10 (metered) | By default a TikTok video is read from its cover frame. Send `frames=all` to read the whole video. The answer then adds a `timeline` with the second each block of text first appears. The [on-screen text](/docs/recipes/on-screen-text.md) cookbook shows a full run. ## When there is no transcript A video with no captions and no speech, or one that was deleted, returns `404 RESOURCE_NOT_FOUND` and is refunded. The error says why, and [error handling](/docs/errors.md#resource-not-found) lists each reason. A video longer than AI transcription covers is refunded too. The [section below](#transcripts-cost-the-same-at-any-length) gives the length limits and the price of each platform. ## Transcripts cost the same at any length Each transcript call charges a flat price when it succeeds, whatever the video's length: | Endpoint | Price | | --- | --- | | `/v1/tiktok/post/transcript` | 10 credits | | `/v1/instagram/media/transcript` | 10 credits | | `/v1/facebook/post/transcript` | 10 credits | | `/v1/twitter/tweet/transcript` | 10 credits | | `/v1/youtube/video/transcript` | 3 credits | On TikTok, `use_ai_as_fallback=true` transcribes the audio when the video has no caption track, at the same price. AI transcription covers videos up to 10 minutes (600 seconds), and Facebook Ad Library ad transcripts cover up to 2 minutes. For an X video over 10 minutes, `/v1/twitter/tweet/transcript` returns the captions X publishes for it when there are any. A longer video with no such captions is refunded, like a video with no speech or a deleted one. `/v1/youtube/video/subtitles` (1 credit) lists a video's caption tracks, each with its language and a download URL. `/v1/youtube/video/transcript` returns the caption text itself, with timestamps. ## Many videos at once `POST /v1/youtube/transcripts` takes a list of YouTube video ids and returns one row per video. You pay only for the transcripts that came back. [Batch](/docs/batch.md) covers it and the other list endpoints. For another platform, [background jobs](/docs/async.md) run many calls to one endpoint as one job. ## Next steps - [Video transcription](/docs/recipes/video-transcription-api.md): One function for the spoken words of any social video. - [On-screen text](/docs/recipes/on-screen-text.md): Read the text creators put on the video itself. - [Batch](/docs/batch.md): Many transcripts in one call, billed per row.