Transcribe Audio

Turn speech in audio or video files into text — watch the transcript stream in live, then download it as TXT, SRT, VTT, or JSON.

⌁ SERVER JOBAPI

Drop an audio or video file to transcribeMP3, M4A, WAV, OGG, FLAC, MP4, MOV, WebM, MKV — or click to browse. Up to 25 min and 524 MB.

HOW IT WORKS

01
Drop an audio or video file
Drop MP3, M4A, WAV, OGG, FLAC, MP4, MOV, WebM, or MKV onto the page — or click to browse. The upload shows a real progress bar driven by true bytes sent, and the limits are stated up front.
02
Our server transcribes it
A self-hosted Whisper model runs on our own CPU server — no third-party AI APIs, so your file is never shared with another company. If the queue is busy you see your actual position, never a fake spinner.
03
Watch the transcript stream in
Your words appear line by line as they are recognized, with an honest percent and a measured time estimate — progress comes from real segment timestamps, not an animation.
04
Download four formats
Get the finished transcript as plain text (.txt), subtitles (.srt or .vtt), or structured JSON with segment timestamps. Read it inline, toggle timestamps, or copy it straight from the page.

Which file formats are accepted?: Nine common containers: MP3, M4A, WAV, OGG, FLAC audio and MP4, MOV, WebM, MKV video. The server reads the actual media stream, so a broken or unsupported file is rejected with a clear message — never a silent failure.
How long can my file be?: Up to the duration and size limits shown on the page. Those numbers come from the live tool configuration, derived from real benchmarks on our hardware — the page never promises more than the server accepts.
What happens to my file?: Files upload over HTTPS and auto-delete after 24 hours. Your upload is used only to produce your transcript, and the downloadable results expire and are removed on the same 24-hour clock.
Which languages are supported?: Whisper auto-detects the spoken language across dozens of languages. You can also pick the language explicitly before transcribing — useful for short clips, where detection has less audio to work with.
Is it really free?: Yes. No signup, no watermark, no trial clock. OpenToolset is open source and the model runs on our own server — this tool does the whole job for free.

Compress JPEG, PNG, or WebP images with a quality slider — runs entirely in your browser. Files never leave your device.

Remove Backgroundsoon

Cut the background from any image — coming soon.

Files upload over HTTPS and auto-delete after 24 hours.