TranscriptFetch

TranscriptFetch

TranscriptFetch turns a video URL into timestamped, structured text in one call. It reads published captions where they exist and transcribes the audio when they don't, returning the same JSON shape either way — across YouTube, TikTok, Instagram, X, Facebook and direct media files (mp4, mp3, wav). Built for RAG pipelines, research datasets and agents that need primary spoken sources.

Base URL: https://transcriptfetch.com/api/v2/transcripts/ · Docs: transcriptfetch.com/docs · OpenAPI: api/v2/openapi.json · MCP: https://transcriptfetch.com/mcp

How to Get a TranscriptFetch API Key

1. Create an account at transcriptfetch.com/sign-up — no card required.
2. Open Manage API keys and create a key (they look like tf_live_…).
3. Store it server-side as TRANSCRIPTFETCH_API_KEY and send it as a bearer token. Requests without a valid key return 401.

Free tier: 50 credits every month, no card. Failed, blocked or empty results are never billed, and neither is job polling.

First Request

curl https://transcriptfetch.com/api/v2/transcripts/video \
  -H "Authorization: Bearer $TRANSCRIPTFETCH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video":"dQw4w9WgXcQ"}'

The video field accepts a bare 11-character YouTube ID, any YouTube / TikTok / Instagram URL, or a direct media file URL.

{
  "ok": true,
  "request_id": "req_…",
  "data": {
    "kind": "transcript",
    "video_id": "dQw4w9WgXcQ",
    "platform": "youtube",
    "title": "Example video",
    "segments": [
      { "start": 0, "duration": 3.5, "text": "We're no strangers to love" }
    ]
  },
  "usage": { "credits_spent": 1, "balance": 49, "bytes": 14233 }
}

Every response uses the same envelope: { ok, request_id, data, usage }. A synchronous transcript carries either joined text or timestamped segments, depending on the timestamps flag; job and batch results include both.

Endpoints

All under https://transcriptfetch.com/api/v2/transcripts/:

Endpoint	What it does
POST /video	One video → timestamped transcript
POST /batch	Up to 50 videos in one request (500 on Mega and Scale)
POST /search	Find videos by keyword (YouTube, TikTok, Instagram)
POST /channel	Resolve a channel or profile into a paginated video list, newest first
POST /playlist	Playlist listing (YouTube and TikTok)

Every listed row carries a url the transcript endpoint accepts as-is. Instagram profiles use the channel endpoint. v1 (/api/v1/...) is still served for existing integrations under a 12-month deprecation policy — new work should use v2.

When There Are No Captions

mode: "auto" transcribes the audio automatically on the same endpoint. Short media may finish inline within about 45 seconds; longer work returns HTTP 202 with a job_id and poll_url, or you can supply a callback_url to receive the finished result instead of polling.

Their published coverage: captions exist for 82% of YouTube, 41% of TikTok and 38% of Instagram URLs; with audio transcription the API returns text for 99% / 96% / 94% of the same URLs.

Python

import os, requests

res = requests.post(
    "https://transcriptfetch.com/api/v2/transcripts/video",
    headers={"Authorization": f"Bearer {os.environ['TRANSCRIPTFETCH_API_KEY']}"},
    json={"video": "https://youtu.be/dQw4w9WgXcQ", "timestamps": True},
)
data = res.json()
for seg in data["data"]["segments"]:
    print(seg["start"], seg["text"])

JavaScript

const res = await fetch('https://transcriptfetch.com/api/v2/transcripts/video', {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${process.env.TRANSCRIPTFETCH_API_KEY}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({ video: 'https://www.tiktok.com/@user/video/123', timestamps: true }),
});
const { data, usage } = await res.json();
console.log(data.segments.length, 'segments ·', usage.credits_spent, 'credits');

Official Python and Node SDKs read the key from TRANSCRIPTFETCH_API_KEY, and there's a prebuilt n8n community node.

Credits and Pricing

Plan	Price	Credits / month
Free	$0	50 (no card)
Basic	$5	1,000
Pro	$15	10,000
Mega	$45	50,000
Scale	$229	300,000

A successful caption transcript or a YouTube listing call costs 1 credit. AI transcription costs 1 credit per started minute (minimum 1) and is billed only on delivery. Unused plan credits roll over up to 2× the monthly allowance; top-ups never expire. Annual billing is 10 months for 12.

Rate limits: 10 requests/minute on free, 60 on Basic, 300 on Pro, 900 on Mega and Scale.

Error Codes

400 invalid input · 401 invalid key · 402 insufficient credits · 409 idempotency conflict · 422 input must change · 429 respect Retry-After · 500/502/503 server or transient upstream failure.

MCP Server

https://transcriptfetch.com/mcp (streamable HTTP, OAuth or API key) exposes five tools — get_transcript, search_videos, list_channel_videos, list_playlist_videos and get_credits — so Claude, ChatGPT, Cursor, Perplexity and other MCP clients can fetch primary sources themselves with no glue code. A machine-readable server card is published at /.well-known/mcp/server-card.json.

Use Cases

• RAG and AI pipelines — ground answers in what people actually said, with timestamps to cite
• Channel monitoring — poll a channel or playlist with since_video_id; a page with nothing new costs no credits
• Research datasets — normalise interviews, reviews, demos and lectures into one timestamped schema
• Agent tooling — let an assistant retrieve the original spoken source itself over MCP

Read the full API docs →