Whisper speech-to-text for any video, audio or podcast URL: full transcript with timestamped segments plus SRT/VTT subtitles, 99+ languages auto-detected. Works on YouTube and TikTok links, podcast RSS feeds and direct media files with no API key needed.
All notable, user-facing capabilities of the Video & Audio Transcriber. This is
a feature log rather than a strict per-release history.
Unreleased
YouTube videos that YouTube briefly refuses with a "confirm you're not a bot" check are now retried on fresh connections while time remains, instead of returning no transcript.
Long YouTube videos (over about 20 minutes) no longer fail with a download timeout: a video that fits within your minute cap is now fetched in one fast pass, and a longer one gets enough time to download the capped part.
A media download that stops making progress is now abandoned and retried another way instead of using up the whole run, so short run timeouts still get a transcript.
The public proxy picker is gone: the actor chooses its own connection. Your own proxyUrls are still honoured; other proxy settings in saved inputs are ignored.
[0.1.0] — initial release
Any URL to transcript. YouTube, TikTok, Instagram, Facebook, X, Rumble,
SoundCloud, Dailymotion, Twitch and 1,800+ other sites, podcast RSS/Atom feeds
and direct media files (mp3, mp4, wav, m4a, flac, ogg, webm, mov).
Full transcript, timestamped segments, SRT and VTT. Subtitle strings in
the record plus downloadable .srt / .vtt files.
99+ languages auto-detected, optional forced language, optional English
translation, optional word-level timestamps.
Podcast feeds expand to episodes. Point it at an RSS feed and the newest
maxEpisodesPerFeed episodes become one row each.
Bounded cost.maxMinutesPerItem caps what is transcribed per URL, only
transcribed minutes are billed, failed items are never charged, and the run
stops cleanly at the run's maximum cost instead of failing.
Long media is chunked, so a one-hour episode does not run for an hour.
Cookies support for YouTube's sign-in gate and restricted content.