Video-to-Text

Video transcription API with Whisper

Q: Can I transcribe a YouTube video without downloading it?

Yes. Send the video link as source_url. The same works for X posts and Spaces, Twitch, Kick and TikTok. Uploaded files are limited to 20 MB for audio and 50 MB for video, so for long recordings a link is easier.

Q: Does it return timestamps for subtitles?

Yes. Set include_ts to true. Segment timestamps are included in the price; ts_level: "word" adds 50%. Each segment has a start and end time in seconds, which you can turn into SRT or VTT lines. The API does not return subtitle files itself.

Q: Can it tell speakers apart?

Yes. Set diarize to true and each segment gets a speaker label such as SPEAKER_00. It adds 50% to the price.

Upload an audio or video file, or send a link from YouTube, TikTok, X, Twitch or Kick. With WhisperLargeV3Ct2 you get back JSON with the text, the detected language and timestamped segments.

An hour from a file or YouTube costs $0.052, 10 minutes $0.013
Segment timestamps at no extra cost, word-level for 50% more
Send a video URL instead of downloading the file

Get $5 credits See supported sources

No credit card · Free credits valid for 90 days · Then top up and pay as you go, no subscription

WhisperLargeV3Ct2 Video-to-Text

Transcript returned by the API Thanks for joining today's call. We will ship the new pricing page on Monday, and the support team will update the Help Center by Wednesday.

5.2 s file · language detected: en (99.95%) · segment timestamps Cost on deAPI: $0.0051 (price as of October 2026)

What you can transcribe

One endpoint, POST /api/v2/audio/transcriptions, takes either a link or a file.

Timestamps: set include_ts. ts_level segment is included in the price; word adds 50%. The API returns start and end times in seconds, and you convert them to SRT or VTT yourself.
Speakers: diarize labels each segment, and each word when you use word timestamps, with a speaker (SPEAKER_00, SPEAKER_01…) and adds 50% to the price. Word timestamps and speakers together still add 50% once.
Language: the API detects it and returns it with language_probability, or you set it with lang (for example "lang": "id").

Source	How to send it	Example use
Audio and video files	`source_file` (multipart upload, up to 20 MB audio or 50 MB video)	Calls, podcasts, interviews, lectures
YouTube	`source_url`	Subtitles, search over a channel, summaries
TikTok	`source_url`	Captions, social listening, content moderation
X posts and Spaces	`source_url`	Social listening, turning recorded Spaces into text
Twitch and Kick	`source_url`	Stream highlights, sponsor mentions

Transcription API pricing

Billed per request by length: $0.0058 for a 1-minute file, $0.052 for an hour. No subscription.

Length	Price	Files for $5
1 minute	$0.0058	≈ 864
10 minutes	$0.013	≈ 390
1 hour	$0.052	≈ 96

What the free account includes

$5 of credits, about 96 hours of audio from files or YouTube
1 transcription request per minute and 10 a day until your first top-up
No credit card, credits valid for 90 days

Get $5 credits

Prices for files and YouTube links on WhisperLargeV3Ct2 from the deAPI price API, October 2026, in USD. Links from X, Twitch and Kick cost more per hour than files and each platform has its own rate. TikTok is priced by length band. Word timestamps or diarize add 50%. For a quote, send duration_seconds and platform (youtube, twitter, twitch, kick or tiktok) to POST /api/v2/audio/transcriptions/price.

Transcription API example

Send a link as JSON or upload a file as multipart. Replace YOUR_API_KEY with a key from your dashboard.

Endpoint: POST /api/v2/audio/transcriptions
Model ID: WhisperLargeV3Ct2
Input: source_url or source_file (20 MB audio, 50 MB video)
Output: JSON: text, language, segments[] with start/end
Free account limits: 1 request per minute, 10 per day

curl -X POST 'https://api.deapi.ai/api/v2/audio/transcriptions' \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Accept: application/json' \
  -F 'model=WhisperLargeV3Ct2' \
  -F 'include_ts=true' \
  -F 'ts_level=segment' \
  -F '[email protected]'

import time, requests

API = "https://api.deapi.ai/api/v2"
HEADERS = {"Authorization": "Bearer YOUR_API_KEY", "Accept": "application/json"}

job = requests.post(f"{API}/audio/transcriptions", headers=HEADERS, json={
    "model": "WhisperLargeV3Ct2",
    "source_url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
    "include_ts": True,
    "ts_level": "segment",
}).json()["data"]

while True:
    status = requests.get(f"{API}/jobs/{job['request_id']}", headers=HEADERS).json()["data"]
    if status["status"] in ("done", "error"):
        break
    time.sleep(3)

# done: the JSON transcript (text, language, segments) is at status["result_url"]
print(status.get("result_url"))

Full API reference Model page

What is a transcription API?

A transcription API turns speech in an audio or video file into text you can store, search or caption with. deAPI runs OpenAI's open Whisper model (WhisperLargeV3Ct2) on a distributed GPU network and accepts links, so you can send a YouTube URL instead of downloading the video and uploading it.

Create a free account

Sign up and get $5 of credits. No credit card.
Copy your API key

Create a key in the dashboard, or test a link in the Playground.
Send a file or link

POST to /audio/transcriptions and read the transcript from the job.

Transcription API FAQ

An hour of audio or video from a file or a YouTube link costs $0.052, 10 minutes $0.013 and a 1-minute clip $0.0058 (October 2026). Links from X, Twitch, Kick and TikTok each have their own rate, so get a quote from the price endpoint with platform set. You pay per request, with no subscription.

Yes. New accounts get $5 of credits without a credit card, enough for about 96 hours of audio from files. Until the first top-up, an account can send 1 transcription request per minute and 10 a day.

Yes. Send the video link as source_url. The same works for X posts and Spaces, Twitch, Kick and TikTok. Uploaded files are limited to 20 MB for audio and 50 MB for video, so for long recordings a link is easier.

Yes. Set include_ts to true. Segment timestamps are included in the price; ts_level: "word" adds 50%. Each segment has a start and end time in seconds, which you can turn into SRT or VTT lines. The API does not return subtitle files itself.

Whisper is a multilingual model. The API detects the spoken language and returns it with a probability, or you can set it with lang, for example "pl".

Yes. Set diarize to true and each segment gets a speaker label such as SPEAKER_00. It adds 50% to the price.

deAPI runs OpenAI's open Whisper model on its own GPU network. The native API has its own request format. If you already use the OpenAI SDK, deAPI also has an OpenAI-compatible endpoint at https://oai.deapi.ai/v1/audio/transcriptions. It takes file uploads and deAPI model IDs (there is no whisper-1 alias).

Transcribe your first hour with $5 of free credits

Sign up with e-mail or Google. No credit card, no subscription.

Get $5 credits Docs

Video transcription API with Whisper

What you can transcribe

Transcription API pricing

What the free account includes

Transcription API example

What is a transcription API?

Create a free account

Copy your API key

Send a file or link

Transcription API FAQ

How much does the transcription API cost?

Can I try it for free?

Can I transcribe a YouTube video without downloading it?

Does it return timestamps for subtitles?

Which languages does it support?

Can it tell speakers apart?

Is Whisper on deAPI the same as the OpenAI Whisper API?

Transcribe your first hour with $5 of free credits

More on deAPI