Whisper transcription without a server

Bulk speech to text with Whisper (faster-whisper), started from the Apify Console, the API, a schedule or an AI assistant.

Run it on the Apify Store

The job

Send a list of direct audio or video URLs and get one row per file: status (ok, failed with the reason, or too_long), detected language, duration, full text and timed segments. A dead link or a non-audio file marks that row failed and the run goes on with the next file; maxAudioMinutes caps the total audio per run.

Example input

{
  "files": [
    "https://example.com/episode-1.mp3",
    "https://example.com/episode-2.wav"
  ],
  "model": "small",
  "outputFormats": ["segments"]
}

Example output

{
  "url": "https://example.com/episode-1.mp3",
  "status": "ok",
  "language": "en",
  "durationSec": 1820.0,
  "text": "...",
  "segments": [{"start": 0.0, "end": 4.2, "text": "..."}]
}

Illustrative rows showing the output shape, not a real run.

Good to know

• CPU only, so long files are slow; a run that hits the Actor timeout or runs out of memory ends, only per-file problems are caught as failed rows.
• Weakest on noisy audio and on non-English speech with the base model.
• No speaker diarization.

Use it from an AI assistant

The same tool is callable through Apify's MCP server: https://mcp.apify.com/?tools=venture_ace/audio-transcription

Price

No tool fee: you pay only Apify's platform usage for your run.

More about this tool on the main page or the Store listing. A bug or a feature you need? Open an issue on the Store page.