Long Audio Transcription
Transcribe hours-long audio & video — podcasts, meetings, lectures — with cloud Whisper. Files are split in your browser and sent in pieces. Export TXT, SRT, VTT. Credits, no signup.
published
- [CREDITS]
- [NO_SIGNUP]
- [USES_AI]
⚠ This is a Pro tool — it runs server work the browser can't and spends credits. No signup: every visitor gets a daily free credit allowance, and you can top up with a no-signup credit key. The site's free tools stay free. See pricing and the privacy page.
Built for the files the free tool can’t handle
The site’s free Audio to Text tool runs Whisper entirely in your browser — perfect for short, private clips. But a two-hour podcast or a full team meeting is too much for on-device transcription on most machines. This tool is the heavier-duty option: it splits your file into short pieces right in the browser (using ffmpeg.wasm), sends each piece to Cloudflare Workers AI (whisper-large-v3-turbo), and stitches the results back into one timeline so timestamps line up across the whole recording.
Because it uses paid server compute, it runs on credits — about 1 credit per minute of audio. There’s still no signup: everyone gets a free daily batch, and heavier use draws from a credit key you buy and store in your browser.
Why credits, and how to stay private
On-device transcription is free because it uses your hardware. Cloud transcription costs real money per minute, so it can’t be unlimited-free without either ads, accounts, or going broke — none of which fit this site. Credits keep it sustainable and honest: you pay only for what you transcribe, with no subscription and no account.
If a recording is confidential, prefer the free on-device tool — it never uploads anything. This Pro tool sends audio chunks to the model to do the work and keeps none of it, but a server is still involved, so it’s the right trade-off only when you value speed/scale over absolute privacy.
Tips
- Pick the language instead of auto-detect when you know it — it’s faster and a little more accurate.
- Translate to English transcribes any supported language straight into English in one pass.
- Keep the tab open while it works; the splitting and chunk uploads happen in your browser.
- Need subtitles? The SRT and VTT exports are ready to drop into a video editor or upload alongside your video.
Frequently asked questions
How is this different from the free Audio to Text tool?
The free Audio to Text tool runs Whisper entirely on your device (nothing uploaded) and is ideal for short clips and full privacy, but it is limited by your hardware and the on-device model size. This Pro tool is built for long files — a 2-hour podcast or a full meeting. It splits the file in your browser and sends the pieces to cloud Whisper (large-v3-turbo) for fast, accurate results without tying up your CPU. It costs credits because it uses paid server compute. If your file is short or sensitive, use the free on-device tool instead.
What does it cost? Do I need an account?
No account, ever. Pricing is in credits: roughly 1 credit per minute of audio. Every visitor gets a batch of free credits each day (no signup) — enough to try it — and beyond that you spend a credit key you can buy on the pricing page. The credit key is a secret you store in your browser like an API key; there is no email or login. You see the estimated cost before you start.
Is my audio private?
This is a server tool, so it is not as private as the on-device option. Your file is split locally and each chunk is sent to Cloudflare Workers AI to be transcribed, then discarded — it is not stored by this tool. Like any web host, Cloudflare may log standard request metadata. For confidential recordings, use the free on-device Audio to Text tool, which never uploads anything.
Is there a file size or length limit?
No hard limit on the source file. It is split into short chunks in your browser (about 10 minutes each, re-encoded small), so length is bounded by your daily/purchased credits and by keeping the browser tab open while it works. Multi-hour files are fine; very large videos take longer to split.
Which formats work?
Almost any audio or video file, because the in-browser splitter (ffmpeg.wasm) decodes it rather than relying on what your browser natively supports. Audio: MP3, WAV, M4A/M4B, AAC, FLAC, OGG/Opus, WMA, AIFF, ALAC, AMR, AC3, APE, WavPack, AU, Speex and more. Video (the audio track is used): MP4, MOV, WebM, MKV, AVI, FLV, WMV, MPEG, TS, 3GP. If a file has an unusual or missing extension it still works — ffmpeg detects the format from the content. Truly exotic or DRM-protected files are the only ones that may fail.
What if I run out of credits mid-file?
It stops cleanly and gives you the transcript collected so far, with a note to top up. Nothing is lost — you keep the partial result, and buying more credits lets you redo the remainder.
Can it transcribe a YouTube or social video from a URL?
No. Pulling media from those platforms violates their terms and browsers block it anyway. Download your own file first (e.g. YouTube Studio for your own videos), then transcribe it here.