Audio to Text
Turn speech into text with AI — entirely in your browser. Pick a recording, choose a model size, and get a transcript you can copy or download as TXT, SRT or VTT subtitles. Your audio is never uploaded: the speech model is downloaded once and runs on your own device, so there are no accounts, no minute limits and no files on anyone's server.
First use downloads a 77 MB speech model, then your browser caches it.
More transcription tools
The same private, in-browser engine, tuned to a specific job:
How to convert audio to text
- Drop in an audio or video file — MP3, WAV, M4A, FLAC and most recordings work.
- Choose a model size and the spoken language (or leave auto-detect), then click Transcribe.
- Watch the text stream in, then copy it or download it as TXT, SRT or VTT.
About this audio-to-text converter
This tool runs OpenAI's open-source Whisper speech-recognition model directly in your browser. The first run downloads the model you pick (41–249 MB) and caches it, so later transcriptions start without the wait. Nothing about your recording is stored, tracked or sent to a server.
Three model sizes let you trade speed for accuracy: Fast for quick notes, Balanced for most recordings, and Precise when every word matters. Timestamped segments come out as SRT or VTT subtitle files that drop straight into video editors and players, and the translate option turns speech in any supported language into English text.
Frequently asked questions
Is my recording uploaded anywhere?
No. The transcription runs entirely in your browser using a local AI speech model (OpenAI's open-source Whisper). The only download is the model itself — your audio never leaves your device, which you can verify in your browser's network panel.
Which languages are supported?
The model is multilingual and recognizes speech in about 99 languages, including English, Spanish, Chinese, French, German, Portuguese, Japanese and Indonesian. Leave the language on auto-detect, or set it explicitly for slightly better accuracy. You can also translate speech from any supported language directly into English text.
How accurate is it?
Very usable on clear speech, and the larger models are noticeably better. Like all speech recognition it struggles with heavy background noise, strong accents, overlapping speakers and music, and it can invent text during long silences — always skim the result before publishing it.
How long does transcription take?
It depends on your device and the model size: the recording is processed locally, typically a few times faster than real time on a modern laptop with the Balanced model. Longer files simply take longer — keep the tab open while it runs.