Speech to Text API: Audio to Text & Video to Text, Whisper
Pricing
from $8.50 / 1,000 basic audio minutes
Audio transcription and video transcription API: transcribe audio, video, MP3, MP4, podcasts to text with Whisper speech to text. Podcast transcription from RSS feeds, meeting and interview transcription, voice to text, SRT/VTT subtitles, RAG chunks. From $0.01 per audio minute; silence is free.
YOUTUBE_BLOCKED, a
plain message and useInstead pointing to YouTube Transcript Scraper. After the first blocked
YouTube link, the other YouTube links in the batch get the same row at once, with no download
attempt and no compute, so a list of YouTube links no longer trips maxFailedItems.maxFailedItems, the failed-work allowance or an unsupported protocol now ends
SUCCEEDED when every error was caused by the submitted links (bad link, no speech, too long, and
similar). The status message says it stopped early and how many links were not tried. Any Actor
side fault in the batch still fails the run.
No change to prices, charging, recognition or output formats.maxDurationMinutes now defaults to 300
(was 180) and the default run timeout is 2 hours (was 1). Billing is unchanged: per audio minute.
Saved Tasks and API calls that set their own values keep them.
No change to prices, charging, recognition or output formats.Found by an end to end test with real podcast inputs:
README-only update: shorter Store listing with prices per hour and job cost examples. No change to prices, charging, recognition, output formats or defaults.
monitorKey is now called monitorName. Saved Tasks and API calls that
still send monitorKey keep working and resume exactly the same checkpoint; nothing to change.README-only update.
SUMMARY.inputCausedFailureOnly records when this applied.model: tiny inputs still work; Tiny is not a fourth tier.In-flight resource checks, bounded feed/download/SDK operations, lower-overhead decoded sample handling and private recovery checkpoints. Use Resurrect on the original run; reconcile uncertain delivery or billing before retrying. These protections do not guarantee zero loss.
Effective-price preflight and billing receipts, bounded workers, signed file links, timestamp diagnostics, opt-in sentence-aware RAG, paragraphs and updated onboarding.
Podcast RSS/Atom input and repeat-run checkpoints; Markdown and RAG exports.
Five-minute audio processing, no-speech failure handling and cached models. The historical 75-minute fixture checked duration acceptance, not representative long-podcast accuracy.
Basic $0.035, Quality $0.06 and Max $0.12 per rounded-up audio minute. Standard plan discounts apply. Actor-start remains $0.00005 per GB. Before that date every model billed the flat $0.04 headline minute rate; old pinned builds retain their legacy event. See the Pricing tab for applicable rates.