Is my recording sent to OpenAI?

No. The open Whisper weights are downloaded to your browser and run locally. The audio never leaves your device.

Why is the first run slower?

This tool needs a processing engine that is downloaded once and then cached: ffmpeg.wasm: about 30 MB, fetched once from the jsDelivr CDN and cached by your browser · Transformers.js: model weights from the Hugging Face Hub on first use (tens of MB), then cached. Only the engine is downloaded. Your file never leaves your device.

Are my files uploaded?

No. This tool runs entirely in your browser tab. Files are read from your disk into memory and the result is generated locally.

Is it free? Are there limits or watermarks?

Free and open source (MIT). There are no file-size caps, daily quotas, accounts or watermarks.

Does it work offline?

Yes, once the page (and any engine it needs) has loaded, it keeps working without a connection thanks to the service worker cache.

Home / Audio / Audio Transcriber

Audio Transcriber

Transcribe speech with on-device Whisper AI and export SRT.

Free, no uploads, no sign-up. Runs 100% in your browser.

How Audio Transcriber works

Audio Transcriber decodes your audio to 16 kHz mono, then runs OpenAI’s Whisper model (tiny or base) on-device with Transformers.js in a Web Worker, using WebGPU when available. It outputs plain text and timestamped SRT/VTT subtitles.

Input: MP3, WAV, OGG, AAC, FLAC, M4A.

How to use Audio Transcriber

Drop an audio file onto Audio Transcriber (or start recording).
Adjust the options and preview playback.
Process the audio.
Download in the format you picked.

Tips

The tiny model (~40 MB) is fast; base (~80 MB) is more accurate with accents and noise.
Pick the spoken language if you know it. Auto-detect is slower.

Audio Transcriber FAQ

Is my recording sent to OpenAI?: No. The open Whisper weights are downloaded to your browser and run locally. The audio never leaves your device.
Why is the first run slower?: This tool needs a processing engine that is downloaded once and then cached: ffmpeg.wasm: about 30 MB, fetched once from the jsDelivr CDN and cached by your browser · Transformers.js: model weights from the Hugging Face Hub on first use (tens of MB), then cached. Only the engine is downloaded. Your file never leaves your device.
Are my files uploaded?: No. This tool runs entirely in your browser tab. Files are read from your disk into memory and the result is generated locally.
Is it free? Are there limits or watermarks?: Free and open source (MIT). There are no file-size caps, daily quotas, accounts or watermarks.
Does it work offline?: Yes, once the page (and any engine it needs) has loaded, it keeps working without a connection thanks to the service worker cache.

Related tools

Voice Recorder: Record your voice with live waveform and export.
Extract Audio: Pull the soundtrack from a video as MP3, WAV or AAC.
Audio Converter: Convert MP3, WAV, OGG, AAC, FLAC and M4A.
Audio Editor & Trimmer: Trim, fade in/out, change volume and export with a waveform.