GitHubStart fetching
Reference

YouTube Search API

One POST resolves a keyword search into a page of YouTube videos, each with a url the transcript and batch endpoints accept as-is. YouTube is the default platform, and it is the one platform where the search takes filters: upload date, duration, captions only, and sort by relevance or views. type finds channels or playlists instead of videos. No YouTube Data API key, quota or OAuth is involved.

1 credit per page of resultsView pricing

Each page of results is one credit, however many videos it lists; pass next_cursor back as cursor for the next page. Failed requests are free.

On this platform

  • upload_date, duration, captions and sort filter and order YouTube video results; type channel or playlist returns a channel_list or playlist_list instead.
  • captions true keeps only videos with a caption track, whose transcripts will come from captions rather than audio transcription.
  • One page costs 1 credit; pass next_cursor back as cursor for the next page.
POST/api/v2/transcripts/searchidempotent

Search for videos, channels or playlists

Resolve a keyword search into a paginated list of videos (metadata only). Searches YouTube by default; set platform to search TikTok or Instagram instead. Every video row carries a url the transcript endpoint accepts unchanged, plus duration and publish time where the source provides them. On YouTube, upload_date, duration and captions filter the videos and sort orders them, and type finds channels (data.kind channel_list) or playlists (data.kind playlist_list) instead, whose rows' urls the channel and playlist endpoints accept as-is.

Body parameters

querystringrequired
Keyword search query.
Example
platformstring (enum)optional
Where to search: youtube (default), tiktok, or instagram. Every video result's url is accepted by the transcript and batch endpoints as-is. The type, upload_date, duration, sort and captions options search YouTube only.
Example
type"video" | "channel" | "playlist"optional
YouTube only. What to search for: video (the default), channel or playlist. Channels come back as data.kind channel_list, each row's url accepted by the channel endpoint; playlists as data.kind playlist_list, each row's url accepted by the playlist endpoint. A channel row carries the @handle, subscriber count and description; its videoCount is null, because YouTube's channel results no longer show one.
Example
upload_date"hour" | "today" | "week" | "month" | "year"optional
YouTube video search only. Keep videos uploaded within the last hour, today, this week, this month or this year.
duration"short" | "medium" | "long"optional
YouTube video search only. Keep short (under 4 minutes), medium (4 to 20 minutes) or long (over 20 minutes) videos.
sort"relevance" | "views"optional
YouTube only. Result order: relevance (the default) or views (most viewed first). YouTube search no longer sorts by upload date or rating, so neither is offered; for recent videos, filter with upload_date.
Example
captionsbooleanoptional
YouTube video search only. true keeps only videos with subtitles or closed captions, whose transcripts come from captions rather than audio transcription. Defaults to false: no filter.
Example
limitinteger (1–50)optional
Max items to return per page: videos, or the playlists or channels some listing options return. Defaults to 5.
Example
cursorstringoptional
Opaque pagination cursor from a previous response's next_cursor (max 256 characters). Omit for the first page. Cursors are scoped to the listing that issued them, so send the same platform and listing options with each page. Offset-based sources cannot page past the first 2000 items.

Request example

curl https://transcriptfetch.com/api/v2/transcripts/search \
  -H "Authorization: Bearer $TRANSCRIPTFETCH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"how transformers work","limit":10}'

Responses

SuccessExample response envelope
{
  "ok": true,
  "request_id": "req_…",
  "data": {
    "kind": "video_list",
    "source": "search",
    "platform": "youtube",
    "videos": [
      {
        "videoId": "Hq3Lz8pVt2K",
        "url": "https://www.youtube.com/watch?v=Hq3Lz8pVt2K",
        "title": "Transformers, explained step by step",
        "duration": 1580,
        "channel": "Example Channel",
        "publishedAt": "2026-09-08T00:00:00Z",
        "stats": {
          "plays": 184000
        }
      }
    ],
    "next_cursor": "eyJvIjoxMH0"
  },
  "usage": {
    "credits_spent": 1,
    "balance": 656,
    "bytes": 0
  }
}
Full endpoint referenceEvery URL format this platform accepts, with an example of eachProduct page

Every YouTube endpoint

YouTube product pageYouTube URL formatsEvery endpoint, all platforms

Frequently asked questions

Is this the YouTube Data API's search.list?
No. It is TranscriptFetch's own search endpoint: no Google project, API key, OAuth or daily quota. One page of results costs 1 credit, and every row's url can go straight to the transcript endpoint.
Which filters does YouTube search support?
upload_date (hour, today, week, month, year), duration (short, medium, long), captions (true keeps only captioned videos) and sort (relevance or views). These are YouTube-only; TikTok and Instagram search take the query alone.
Can I search for channels or playlists rather than videos?
Yes. Send type channel or type playlist. The response's data.kind becomes channel_list or playlist_list, and each row's url is accepted by the channel or playlist endpoint as-is.
How do I page through results?
Each response carries data.next_cursor. Send it back as cursor to get the following page; limit sets the page size, up to 50. Stop when next_cursor is null.