Whisper Audio & Video Transcriber (Speech to Text, SRT/VTT)

What does Whisper Audio & Video Transcriber do?

Whisper Audio & Video Transcriber converts speech in audio and video files into text using OpenAI Whisper (the open-source model, run with the optimized faster-whisper engine). Give it one or many file URLs and get back:

• 📝 the full transcript as plain text
• ⏱️ timestamped segments (and optional word-level timestamps)
• 🎬 ready-to-use SRT and VTT subtitles, plus downloadable .srt, .vtt and .txt files
• 🌍 the detected language (99 languages supported) or an English translation

No API keys, no OpenAI account and no setup: paste a link and run. Works with mp3, wav, m4a, aac, ogg, opus, flac, mp4, mov, webm, mkv and most other audio/video formats. Public Google Drive, Dropbox and OneDrive share links are converted to direct downloads automatically.

Why use this transcriber?

• 💸 Cheap and predictable: from $0.006 per audio minute, the same as the OpenAI Whisper API, with no monthly fee. A 1-hour podcast costs $0.36.
• 🔒 Private: files are processed inside your Apify run and are not sent to any third-party AI API.
• 🧾 You only pay for successful files: broken links, web pages and unreadable files are not charged.
• 🔁 Built for automation: call it from the API, schedule it, or plug it into n8n, Make, Zapier, LangChain or your AI agent through the Apify MCP server.
• 🎞️ Subtitles out of the box: SRT/VTT files you can upload straight to YouTube, Vimeo, Premiere or CapCut.

Use cases

• Transcribe podcasts, interviews, webinars, meetings, lectures and voice notes
• Generate subtitles / closed captions for videos and social media clips
• Turn video content into text for blog posts, SEO, summaries and LLM / RAG pipelines
• Translate foreign-language audio into English text
• Search and analyze call recordings and customer interviews at scale

How much does it cost to transcribe audio?

You pay per started minute of audio (each file's duration is rounded up to the next minute). There is no charge for files that fail.

Model	Best for	Price per audio minute	1 hour of audio
base (default)	Clear speech: podcasts, meetings, voice notes	$0.006	$0.36
small	Accents, background noise, names, non-English audio	$0.015	$0.90

Platform usage is included in the price. Apify's free plan gives you monthly credits, enough to transcribe several hours of audio for free.

How to transcribe audio or video files

1. Click Try for free / Start.
2. Paste one or more file links into Audio or video URLs.
3. Optionally choose the model, set the language (or keep auto) and enable Translate to English.
4. Click Start and wait. A 1-hour file usually takes 3 to 6 minutes with the base model.
5. Download the results as JSON, CSV, Excel or HTML from the Output tab, or grab the .srt/.vtt/.txt files from the run's key-value store.

💡 Tip: give the run more memory (8 GB or 16 GB) to transcribe long files faster. The price per minute stays the same.

Input

{
    "audioUrls": [
        "https://example.com/podcast-episode-42.mp3",
        "https://drive.google.com/file/d/1AbCdEfGhIjKlMnOp/view?usp=sharing"
    ],
    "model": "base",
    "language": "auto",
    "translateToEnglish": false,
    "includeSegments": true,
    "includeSubtitles": true,
    "wordTimestamps": false,
    "initialPrompt": "Apify, Kubernetes, Dr. Nakamura"
}

See the Input tab for every option, including safety limits on file duration and size.

Output

Each file produces one item in the dataset:

{
    "url": "https://example.com/podcast-episode-42.mp3",
    "status": "succeeded",
    "model": "base",
    "language": "en",
    "languageProbability": 0.98,
    "durationSeconds": 1834.2,
    "billableMinutes": 31,
    "text": "Welcome back to the show. Today we're talking about...",
    "wordCount": 5120,
    "segments": [
        { "start": 0.0, "end": 3.4, "text": "Welcome back to the show." },
        { "start": 3.4, "end": 7.9, "text": "Today we're talking about..." }
    ],
    "srt": "1\n00:00:00,000 --> 00:00:03,400\nWelcome back to the show.\n...",
    "vtt": "WEBVTT\n\n00:00:00.000 --> 00:00:03.400\nWelcome back to the show.\n...",
    "srtUrl": "https://api.apify.com/v2/key-value-stores/.../records/transcript-1.srt",
    "vttUrl": "https://api.apify.com/v2/key-value-stores/.../records/transcript-1.vtt",
    "txtUrl": "https://api.apify.com/v2/key-value-stores/.../records/transcript-1.txt"
}

Failed files are returned with "status": "failed" and a human-readable error, so a batch never stops because of one bad link.

Use it from code, n8n or AI agents

Python

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("cyuli/whisper-transcriber").call(run_input={"audioUrls": ["https://example.com/talk.mp3"]})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["text"])

JavaScript

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: '<YOUR_APIFY_TOKEN>' });
const run = await client.actor('cyuli/whisper-transcriber').call({ audioUrls: ['https://example.com/talk.mp3'] });
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items[0].text);

No-code: use the Apify integration in n8n, Make or Zapier: "Run Actor" with the input above, then "Get dataset items". AI agents (Claude, ChatGPT, Cursor...) can call this Actor directly through the Apify MCP server.

FAQ

Which languages are supported? All 99 Whisper languages, including English, Spanish, Portuguese, French, German, Italian, Dutch, Russian, Arabic, Hindi, Japanese, Korean and Chinese. Language is detected automatically, or you can set it with an ISO code like es or pt.

Can I transcribe YouTube, TikTok or Instagram links? This Actor needs a direct link to a media file. For YouTube videos that already have captions, a YouTube transcript Actor is cheaper. For other platforms, download the media first (many Apify Actors do this) and pass the file URL.

How accurate is it? Whisper is one of the most accurate open speech recognition models. base works well for clear speech; switch to small for noisy audio, strong accents or uncommon names. Use the Vocabulary hint to help it spell brand names and jargon.

Is there a file size or length limit? By default files up to 2 GB and 4 hours are accepted. You can change both limits in the input. Files above the limits are skipped and not charged.

Does it identify speakers (diarization)? Not yet. Each segment has timestamps but no speaker label.

Is my data safe? Files are downloaded into your own Apify run, transcribed locally with the open-source model and deleted when the run ends. Results stay in your Apify storage.

Feedback

Found a file that doesn't work or need a feature (speaker labels, more models, other output formats)? Open an issue in the Issues tab.