About Quietscribe

Quietscribe is a set of free tools that turn speech into text: transcripts of recordings, timed subtitles for videos, and a converter for subtitle files.

It was built for creators, students, journalists and anyone who needs the words from a recording but doesn't want to upload it to a stranger's server or pay per minute.

How it works

The speech recognition runs inside your own browser. When you choose a file, your browser reads it and an AI model processes it on your device. The first time, your browser downloads the model and keeps it for next time. Your file is never sent to us.

Because the work happens on your device, each file costs us nothing, which is why there are no minute limits, no sign-up and no watermark.

Built on open-source work

The speech recognition model is Whisper, released as open source by OpenAI. It runs in the browser using Transformers.js by Hugging Face. The site is hosted on Cloudflare.

Limits to know about

AI transcription is very good but not perfect. Names, accents, music and people talking over each other can cause mistakes, so check anything important before you publish it. Speed depends on your device: Chrome or Edge on a computer is fastest.

Questions or ideas? See the contact page.