Whisper Audio & Podcast Transcriber (SRT, VTT, Translate)

Transcribe audio and video files, or the latest episodes of a podcast RSS feed, with Whisper running inside this Actor: no third-party API or account needed. You get text, timestamped segments, SRT and VTT subtitles, the detected language, and optional translation to English. Upload files or paste links, including Google Drive and Dropbox share links.

Consent and rights: only transcribe recordings you own or have permission to transcribe, and respect the privacy of the people speaking.

Quick start

1. Click Start. The prefill transcribes a 45-second clip of the Gettysburg Address (LibriVox reading, public domain) with the fast model for $0.016.
2. Add your own files, or a podcast RSS feed together with "Latest episodes per feed".
3. Open the Output tab. The Text only view shows just the transcripts; the full view includes download links for TXT, SRT and VTT.

Models

Quality	Model	Best for	Price per audio minute
fast	Whisper base	Clear speech, drafts, search indexing	$0.006
standard (default)	Whisper small	Most podcasts and meetings, translation	$0.015
accurate	Whisper large-v3-turbo	Accents, noisy audio, names	$0.05

Each successfully transcribed file also costs $0.01, which covers download and model start-up. Examples (standard): a 3-minute clip costs $0.055; a 30-minute episode costs $0.46. With the fast model, a 30-minute episode costs $0.19. Failed files are never charged, and that includes dead links, sign-in pages and web pages that aren't audio. Bronze, Silver and Gold plans get 10/20/30% off.

Long episodes: set Maximum cost per run high enough. If it's too low, the row says how long the recording is and roughly how much it needs, and you aren't charged.

Output

{ "source": "https://api.apify.com/v2/key-value-stores/1jeoFyhuQnpLWO4x9/records/gettysburg-address-40s.mp3", "ok": true,
  "language": "en", "durationSeconds": 45.1, "billedMinutes": 1, "quality": "fast", "charged": true,
  "text": "Four score and seven years ago our fathers brought forth on this continent a new nation ...",
  "segments": [{ "start": 0.0, "end": 6.2, "text": "Four score and seven years ago ..." }],
  "files": { "txt": "https://api.apify.com/...txt", "srt": "https://api.apify.com/...srt", "vtt": "https://api.apify.com/...vtt" } }

Limits

Up to 10 files per run (including RSS episodes), 500 MB and 120 minutes per file, and 4 GB memory. Transcription runs on CPU, so long files take a while; the fast model is the quickest. Inputs are deleted after each file, and audio and transcripts are never logged. Output stays in the run's storage under Apify's normal retention.

FAQ

Which languages? About 99, detected automatically. Set language (for example es) to skip detection. Speaker labels? Not yet. You get one transcript per file, with timestamps. YouTube or TikTok links? They're refused on purpose. Use files or podcast feeds you have rights to. Why "LOGIN_OR_VIEWER_PAGE"? The share link isn't public. Set it to "Anyone with the link" or upload the file. You weren't charged. Can an AI agent call it? Yes, through the Apify API or MCP, and it supports x402 payments.

Integrations: no-code and AI

Zapier: new podcast episode → transcript in Notion

1. Trigger: RSS by Zapier → New Item in Feed (your show's feed).
2. Action: Apify → Run Actor → Whisper Audio & Podcast Transcriber, run asynchronously (start-up alone takes about 40 s, over Zapier's 30 s sync limit). Input: {"audioUrls": ["{{Enclosure URL}}"], "quality": "standard"}.
3. Second Zap: Apify → Finished Actor run (this Actor) → Fetch dataset items → Notion Create Database Item with Title = episodeTitle or source, Content = text, and the SRT link = files.srt.

Make: weekly show → transcript plus YouTube captions

Schedule → Apify Run an Actor with {"rssFeeds": ["https://example.com/feed.xml"], "latestEpisodes": 1} → a separate scenario with Watch Actor Runs → Get Dataset Items → Google Docs Create a Document (text) and HTTP Get a file (files.srt) for captions.

n8n: voice notes → meeting minutes with an LLM

Telegram/Drive trigger → Apify: Run an Actor and get dataset (wait for finish = true) → OpenAI / AI Agent node ("Summarise {{$json.text}} into decisions and action items") → Slack or Email.

LangChain: searchable podcast archive

from langchain_apify import ApifyWrapper
from langchain_core.documents import Document
docs = ApifyWrapper().call_actor(
    actor_id="steady_kahikatea/whisper-transcriber",
    run_input={"rssFeeds": ["https://example.com/feed.xml"], "latestEpisodes": 5, "quality": "fast"},
    dataset_mapping_function=lambda it: Document(page_content=it.get("text") or "",
        metadata={"source": it["source"], "title": it.get("episodeTitle"), "published": it.get("published")}),
).load()

LlamaIndex: one node per timestamped segment

from llama_index.core import Document
from llama_index.readers.apify import ApifyDataset
# Run the Actor once, then load the run's dataset by ID.
docs = ApifyDataset("<APIFY_TOKEN>").load_data(dataset_id="<defaultDatasetId>",
    dataset_mapping_function=lambda it: Document(text=it.get("text") or "", metadata={"source": it["source"]}))

Sample input → output: {"audioUrls": ["https://example.com/ep42.mp3"], "quality": "standard"} → {"ok": true, "language": "en", "durationSeconds": 1802, "billedMinutes": 31, "text": "Welcome back to the show ...", "files": {"srt": "https://api.apify.com/...srt"}, "charged": true}. Cost: $0.475.

Want a finished PDF or Word transcript? See Podcast & Audio to PDF / Word Transcript, which chains this Actor with the document generator.

Tip: Zapier's synchronous Run Actor times out after 30 seconds. For anything longer, run it asynchronously and start a second Zap from the Finished Actor run trigger. Make's synchronous timeout depends on your plan; use Watch Actor Runs for long jobs.

Also works with: the Apify API ("API" tab), the Apify MCP server for Claude/Cursor-style agents, and x402 payments.

Docs used: Zapier · Make · n8n · LangChain · LlamaIndex

For AI agents

Actor: steady_kahikatea/whisper-transcriber
Input: {"audioUrls": ["https://example.com/ep.mp3"], "quality": "fast"}  or  {"rssFeeds": ["https://example.com/feed.xml"], "latestEpisodes": 2}
Output: rows with text, segments[], srt, vtt, language, billedMinutes, charged, error

Licenses

faster-whisper and CTranslate2 (MIT); Whisper models (MIT).

Questions? Use the Issues tab. If it's useful, a rating helps others find it.