Try it
Run a live transcript request
One real call, no signup and no key. Paste a video link and read the transcript JSON we hand back.
Endpoint & Pricing
One Endpoint, Five Price Tiers
One call is one video, including two-hour lectures. The rate falls as your deposit tier rises; there is no monthly plan.
Video transcript
GET /api/social/youtube/transcript—videorequired- Any video id or YouTube link, including Shorts and live
langpicks the language;outputpicks lines or text- Timed lines and full text in one flat response
Pay as you go, no subscription. Growth takes one qualifying $100 deposit, Scale $500, Enterprise $1,000 and Enterprise Plus $1,250 for new credit. The deposit is spendable balance rather than a fee; existing credit keeps its original schedule. Refunded or charged-back amounts don’t count toward your tier. Every account starts with 10 free API calls, shared across all endpoints, and this endpoint can spend them.
Quick Start
A Transcript in One Request
One header and a query string. The answer is a flat object, so the text is a single field read away.
# Default: timed lines plus the full text, in the video's own main language. curl "https://apiserpent.com/api/social/youtube/transcript?video=8jPQjjsBbIc" \ -H "X-API-Key: sk_live_your_api_key" # Any YouTube link works. German transcript, full text only. curl -G "https://apiserpent.com/api/social/youtube/transcript" \ --data-urlencode "video=https://youtu.be/8jPQjjsBbIc" \ --data-urlencode "lang=de" \ --data-urlencode "output=text" \ -H "X-API-Key: sk_live_your_api_key"
import requests API_KEY = 'sk_live_your_api_key' BASE = 'https://apiserpent.com' HEADERS = {'X-API-Key': API_KEY} resp = requests.get( f'{BASE}/api/social/youtube/transcript', headers=HEADERS, params={'video': 'https://www.youtube.com/watch?v=8jPQjjsBbIc', 'lang': 'en'}, timeout=60, ) data = resp.json() print(data['title'], '|', data['language'], '| auto-generated:', data['is_auto_generated']) for seg in data['segments'][:5]: print(f'{seg["start"]:8.2f}s {seg["text"]}')
(async () => { const params = new URLSearchParams({ video: 'xAcTmDO6NTI', output: 'text', }); const res = await fetch( 'https://apiserpent.com/api/social/youtube/transcript?' + params, { headers: { 'X-API-Key': 'sk_live_your_api_key' } } ); const data = await res.json(); console.log(data.title, '-', data.duration_seconds, 'seconds'); console.log(data.text.slice(0, 300)); })();
import requests HEADERS = {'X-API-Key': 'sk_live_your_api_key'} resp = requests.get( 'https://apiserpent.com/api/social/youtube/transcript', headers=HEADERS, params={'video': '8jPQjjsBbIc'}, timeout=60, ) data = resp.json() # Group the timed lines into chunks of about 60 seconds, keeping each chunk's start time. chunks, current, chunk_start = [], [], 0 for seg in data['segments']: if current and seg['start'] - chunk_start >= 60: chunks.append({'start': chunk_start, 'text': ' '.join(current)}) current, chunk_start = [], seg['start'] current.append(seg['text']) if current: chunks.append({'start': chunk_start, 'text': ' '.join(current)}) for ch in chunks[:3]: print(f'[{ch["start"]:.0f}s]', ch['text'][:80]) # Each chunk carries data["video_id"] and its start time, so a hit links straight to the moment.
{
"success": true,
"video_id": "8jPQjjsBbIc",
"title": "How to stay calm when you know you'll be stressed | Daniel Levitin | TED",
"channel": "TED",
"duration_seconds": 740,
"language": "en",
"is_auto_generated": false,
"available_languages": [
{ "code": "ar", "name": "Arabic", "is_auto_generated": false },
{ "code": "en", "name": "English", "is_auto_generated": false },
{ "code": "en", "name": "English (auto-generated)", "is_auto_generated": true }
// ...41 entries in all
],
"segments": [
{ "start": 13.24, "duration": 2.56, "text": "A few years ago, I broke into my own house." },
{ "start": 16.88, "duration": 1.216, "text": "I had just driven home," },
{ "start": 18.12, "duration": 2.536, "text": "it was around midnight in the dead of Montreal winter," },
{ "start": 20.68, "duration": 2.296, "text": "I had been visiting my friend, Jeff, across town," }
// ...
],
"text": "A few years ago, I broke into my own house. I had just driven home, it was around midnight in the dead of Montreal winter, ..."
}
// GET /api/social/youtube/transcript?video=xAcTmDO6NTI&output=text { "success": true, "video_id": "xAcTmDO6NTI", "title": "Lecture 1: Introduction to CS and Programming Using Python", "channel": "MIT OpenCourseWare", "duration_seconds": 3810, "language": "en", "is_auto_generated": false, "available_languages": [ { "code": "en", "name": "English (auto-generated)", "is_auto_generated": true }, { "code": "en", "name": "English - CC (English)", "is_auto_generated": false }, { "code": "es", "name": "Spanish - CC (Spanish)", "is_auto_generated": false } ], "segments": [], "text": "[SQUEAKING] [RUSTLING] [CLICKING] ANA BELL: All right, so welcome to the first lecture of 6.100L. That's our new number. My name is Ana Bell. ..." }
Both samples are real responses recorded on 20 November 2026, trimmed for length. output=text returns the full text with an empty segments list, which keeps a long lecture light. The response shape is the same on every call.
Parameters & Fields
Every Parameter and Field
Three request parameters and nine response fields, all at one flat level and named the same on every call.
Request Parameters
- video (required): id or any YouTube link
- lang: language code such as en, de, pt-BR
- output: segments (default) or text
- Header: X-API-Key
Response Fields
- video_id, title, channel
- duration_seconds, language
- is_auto_generated
- available_languages[] (code, name, is_auto_generated)
- segments[] (start, duration, text)
- text
Links and ids it accepts
The 11-character video id, or a link in any of these forms: watch, youtu.be, Shorts, embed, live and mobile. Paste whatever your users paste.
Language and caption type
Leave lang out for the video’s own main language, with a human-written transcript preferred over an automatic one. Name a code and you get that language; available_languages lists every option.
Timed lines or one string. start and duration are in seconds, so a line maps straight to a moment in the video. With output=text you get only the full text and an empty segments list, the lighter choice for summaries and search indexes.
Why a transcript API
What is a YouTube transcript API, and why teams use it
A transcript is the words of a video as text. Once they are text, you can search, quote, translate and feed them to a model.
Words with timestamps
Each line carries its start time and length in seconds, so a search hit can link to the exact moment, and a quote can be checked against the video.
Or ask for output=text and take the whole transcript as a single string.
Every language the video has
A single video can carry dozens of caption languages. The TED talk in the sample offers 41, English among them in both human-written and automatic form.
Each entry in available_languages says which is which, so you choose before you ask.
Human or automatic, labelled
is_auto_generated tells you whether a person wrote the transcript or YouTube produced it automatically, on the transcript you got and on every language the video lists.
Quality-sensitive work can prefer the human track and treat the rest accordingly.
Captions, not media
The API returns caption text and the basics of the video: title, channel and length. It does not download or store the video itself.
For views, likes and channel stats, the YouTube API sits on the same key.
Price comparison
What YouTube Transcripts Cost, Everywhere
The lowest published rate for a transcript at each vendor we compared, in the same unit, with what it takes to reach it.
| Provider | Unit charged | Cost per 1,000 | What the lowest rate requires | Free tier |
|---|---|---|---|---|
| Serpent | per call, one video | $0.016/1K calls | Enterprise Plus tier, $1,250 deposit that stays as spendable balance. Pay as you go from $0.05/1K with no subscription. | 10 free API calls |
| Apify seemuapps transcript actor | per transcript | $0.10/1K transcripts | Flat pay-per-transcript rate, no volume tiers. Up to 10 videos per run. | Apify free-plan credit |
| Scrapingdog | per request, 5 credits | $0.136/1K requests | $30,000/month plan. Entry rate $1.00/1K on the $40/month plan. | 100 credits (20 transcripts) |
| HasData | per request, 10 credits | $0.693/1K requests | $208 plan with 3M credits. Entry rate $2.45/1K on the $49 plan. | 1,000 credits/month (100 transcripts) |
| Supadata | per transcript, 1 credit | $0.897/1K transcripts | $897 plan with 1,000,000 credits. Entry rate $16.67/1K on the Basic plan. | 100 credits/month |
| ScrapeCreators | per request, 1 credit | $0.994/1K requests | $497 pack with 500,000 credits. Entry rate $1.88/1K on the $47 pack. | 100 credits |
| TranscriptAPI.com | per request, 1 credit | $1.50/1K requests | Top-ups on the $54/year plan. Entry rate $5.00/1K on the monthly plan. | 100 credits |
| Apify johnvc transcript actor | per video | $2.00/1K videos | Apify Gold tier or above; $2.75/1K on the free Apify tier. | Apify free-plan credit |
| youtube-transcript.io | per transcript | $8.33/1K transcripts | Pro plan, $24.99/month for 3,000 transcripts. Entry rate $9.99/1K. | 25 transcripts/month, API not included |
Prices checked 20 November 2026. Promotional trial rates that end within weeks are not counted. Our figure is the Enterprise Plus rate, so every competitor cell shows that vendor’s lowest published rate too, not its entry rate; the entry rate sits in the requirements column. Credit-based plans are converted at the credits each transcript uses on that vendor’s own documentation. Vendors that do not publish what a transcript costs in credits are left out. Our entry rate is $0.05/1K calls.
Use cases
What Teams Build With YouTube Transcripts
Four jobs that account for most transcript traffic, and the parameters each one leans on.
AI, RAG and LLM summaries
Turn a playlist of talks into a searchable knowledge base. Take segments, group them into chunks and embed each one with its start time, so an answer cites the minute it came from.
For a one-shot summary, output=text hands the model the whole transcript in a single field.
Content repurposing
One recording becomes a blog post, a newsletter, a set of social quotes and a captioned clip. The timed lines give you the cut points; the full text gives you the draft.
A long lecture is still one call, so a back catalogue is a loop, not a project.
SEO and market research
Read what competitors and customers actually say on video: product reviews, conference talks, founder interviews. Run the text through keyword counts, topic models or a sentiment pass.
Pair it with channel and video stats from the YouTube API to see which talks drew an audience.
Accessibility and translation
Ask for lang=de, lang=pt-BR or any code in available_languages and ship a transcript next to the video on your own page, in your readers’ language.
is_auto_generated tells your editors which transcripts to proofread first.
Tested on real videos
Measured on 54 Real Videos
We drew the test set before we measured anything, so the results reflect the videos people actually ask for.
Every public video with captions returned its transcript. The set covered 11 languages, Shorts, two-hour lectures, music and live archives. The median call took about 2.4 seconds.
Fixed set of 54 videos chosen before measuring, run 20 November 2026. Median is the time of the answered call.
FAQ
YouTube Transcript API Questions
Start using the YouTube Transcript API
10 free API calls, shared across every free-eligible endpoint. No card, no subscription, no monthly plan. From $0.016 per 1,000 calls.
Try for Free






