Transcribe MP3/WAV/FLAC/OGG/WEBM/M4A audio into text with speaker diarization, timestamps, context-aware terminology, and SRT/VTT. No external transcription API key required. Built for batch/API/MCP workflows.
2026-10-21: Runs without a paid-plan signal no longer fail (0.1)
Fixes
A run that receives no paid-plan signal from the platform (for example the Store's automatic test run) now succeeds with FREE_PLAN_REQUIRES_UPGRADE rows instead of failing. Nothing is transcribed or charged, same as on a free plan.
RUN_SUMMARY now includes userPlanSignal, a safe category of that signal, to help diagnose plan questions. An unreadable signal ends the run as PLAN_CHECK_FAILED before anything is downloaded or charged.
2026-10-19: Transcribe public audio links with speaker labels (0.1)
Features
Paste 1-100 public HTTPS links to audio files (MP3, WAV, FLAC, OGG, WEBM, M4A or AAC). Each link becomes one Dataset row with a status, the transcript, speaker labels when diarization is on, timestamps and SRT/VTT subtitles.
No speech-to-text key is needed: the included transcription costs $0.007 per started audio minute, with platform usage included. With your own Mistral key in mistralApiKey, a minute costs $0.004 here and Mistral bills the transcription to your account.
Runs started by an Apify integration from another Actor or Task are accepted. Put the audio links in audioUrls of the integration input.
Reliability and billing
A file with no detectable speech is NO_SPEECH. Unreachable, undecodable, oversized and over-duration files are explicit per-file failures. None of these is charged.
When the spending limit or the run timeout is reached, finished rows stay in the Dataset and unfinished files are marked and not charged. Runs without a spending limit work as well.
Free Apify plans cannot run this Actor. Set maxTotalChargeUsd on API and agent runs to bound the bill; the lowest accepted limit is $0.008.
This is the first version, so there is no earlier input to migrate.
