# Transcribe Audio in Your Browser — Nothing Uploaded

> Run Whisper locally to turn speech into text, SRT, VTT or a timestamped transcript. Works offline after the first run.

This transcriber runs OpenAI's Whisper model in your browser using WebGPU or WebAssembly. Audio is decoded locally, resampled to 16 kHz and processed in 30-second windows, producing a transcript with segment timings that can be exported as plain text, SRT, WebVTT or JSON. The file is never uploaded, and after the first model download the tool works offline.

**URL:** https://convertto.tech/t/ai-audio-transcription
**Category:** On-Device AI (https://convertto.tech/c/local-ai-tools)
**Privacy:** Runs entirely in the browser; no upload
**Cost:** Free, no sign-up
**Last updated:** 2026-08-01

## Key facts

- **Uploads:** None — the audio never leaves your device
- **Languages:** 99 on every model except tiny, which is English only
- **Best accuracy:** Whisper large-v3-turbo — a 1 GB download, cached after the first run
- **Offline:** Works with no network after the first model download
- **Privacy:** Runs entirely in your browser — nothing is uploaded
- **Cost:** Free, unlimited, no sign-up

## How to use

1. Select your audio or video — the file stays on your device and is never uploaded.
2. Choose the model.
3. Choose the language.
4. Choose the task.
5. Choose the output.
6. Set the characters per subtitle line.
7. Press Run, then download the result when it is ready.

## FAQ

### Is my recording uploaded?

No. The file is decoded in your browser and the model runs on your device. This is the whole reason the tool exists — the recordings people most need transcribed are usually the ones they are not allowed to upload anywhere.

### How long does it take?

With WebGPU, roughly real-time or better on the base model — a ten-minute recording in a few minutes. On the WebAssembly fallback expect several times the audio duration. The first run also includes the model download.

### How accurate is it compared with a paid service?

At the large-v3-turbo setting this runs the same family of model most hosted services do, so on clear speech the gap largely closes. The hosted services still win on diarisation, on very noisy audio, and on turnaround for long files — but the model here is no longer the small end of the family. Move up a size before concluding the audio is the problem.

### Can it tell speakers apart?

No. Whisper transcribes speech but does not do diarisation, so a two-person interview comes back as one continuous transcript. The segment timings make it straightforward to label afterwards.

## Sources

- [Robust Speech Recognition via Large-Scale Weak Supervision (Whisper)](https://arxiv.org/abs/2212.04356) — arXiv
- [WebVTT specification](https://www.w3.org/TR/webvtt1/) — W3C

## Related tools

- [Audio to Subtitles](https://convertto.tech/t/audio-to-subtitles): Generate timed SRT or WebVTT subtitles from an audio recording.
- [Video to Subtitles](https://convertto.tech/t/video-to-subtitles): Generate timed SRT or WebVTT subtitles from a video, in your browser.
- [Audio to Text Transcriber](https://convertto.tech/t/audio-to-text): Transcribe speech in an audio file to text with Whisper, running on your own device.
- [Subtitles to Plain Text](https://convertto.tech/t/subtitles-to-text): Turn an SRT, VTT or ASS subtitle file into a clean transcript.
- [Video to Text Transcriber](https://convertto.tech/t/video-to-text): Pull a transcript out of a video's speech without uploading the file.
- [Subtitle Converter](https://convertto.tech/t/subtitle-converter): Convert between SRT, WebVTT, ASS/SSA, MicroDVD and plain text.
- [AI Text to Speech](https://convertto.tech/t/ai-text-to-speech): Turn text into natural speech with a neural voice running in your browser, and download the audio as a WAV file.
- [Subtitle Merger](https://convertto.tech/t/subtitle-merger): Join two subtitle files end to end, or stack two languages into bilingual cues.