Drop an MP4, MOV, MP3 or WAV and get an accurate AI transcript in minutes. Auto language detection, one-click SRT export, optional summary. No login, no watermark, no app.
If you've spent 20 minutes typing out a single interview, you already know manual transcription is the worst part of video work. This free online transcriber does it in a fraction of the time — upload any video or audio file, and a browser-side compressor strips the audio and streams it to an OpenAI Whisper pipeline. You get back a time-stamped transcript in the same session, ready to copy, download as SRT/VTT, or feed into a summary. No account, no install, and your audio file is auto-deleted within 24 hours.
The underlying engine is Whisper v3, OpenAI's state-of-the-art speech model — roughly 98% word accuracy on clean speech in English, and between 88% and 95% for most of the 98 languages it supports. Accuracy drops with heavy background music, strong crosstalk, or muffled phone audio, but you can fix those segments in-line: the result page shows every timestamped segment individually so you can spot-edit before exporting.
Uploading gives you control and privacy. Platforms like YouTube and Instagram can revoke access to third-party scrapers at any time — if their API changes on a Tuesday, every URL-based transcriber breaks. An uploaded file is yours: if it's already on your device, we can transcribe it. The flip side is file size, which is why we compress to 16 kHz mono MP3 in your browser before uploading — a 2-hour MP4 shrinks to roughly 30 MB, well under the limits.
Search a 90-minute meeting for a single phrase. Generate SRT subtitles for YouTube, Premiere, or DaVinci Resolve. Extract a 5-bullet AI summary when you don't have time to read the whole thing. Translate to Chinese, Spanish, French, German, Japanese, or Portuguese with a single checkbox. Copy the text into Notion, Docs, or Obsidian as a searchable note. The free tier covers 10-minute files × 3 per day, which is enough for most podcasts or lectures; Pro extends to 4 hours per file.
Yes. Three transcriptions per day up to 10 minutes each, with SRT/VTT export and AI summary included. Pro removes the daily limit and extends per-file length to 4 hours.
No. You can transcribe anonymously. We throttle by IP to prevent abuse, so three anonymous jobs per 24 hours.
The compressed audio is stored on Cloudflare R2 just long enough to transcribe it, then auto-deleted within 24 hours. The original video never leaves your device. Transcript text is kept so you can reopen the result page.
Whisper auto-detects among 98+ languages including English, Chinese, Spanish, French, German, Japanese, Portuguese, Arabic, Hindi, Russian, Turkish, Vietnamese, Thai, Korean, and Italian.
Yes — check the 'AI summary' box before uploading. A 5-bullet summary is generated with Claude Haiku (or a Workers AI Llama fallback) once the transcript is ready.
The output is shown segment-by-segment with timestamps, and the TXT/SRT/VTT downloads are plain text — open them in any editor and correct names or tricky terms before publishing.