NoteFast / Speech to text
Speech to text
Press Dictate, talk for up to a minute, press Stop. Your words land in the pad, punctuated, ready to edit.
Each clip is up to 60 s. Record again to keep adding. Works best in a quiet room with the mic near you.
Getting clean dictation
- Say the punctuation only if you want it literal. The model adds commas and full stops from your intonation, so talk in natural sentences.
- Pause, don't um. A short silence reads as a sentence break; fillers sometimes get transcribed.
- One idea per clip. Short clips are faster to check and fix than one long ramble.
- Tidy up after. "Fix grammar" cleans slips without changing your words; "Bullet points" turns a spoken brain-dump into a list.
A comfortable speaking pace is around 150 words per minute – roughly three times faster than most people type while composing. That makes dictation well suited to first drafts, meeting recaps and ideas on the move. See how that converts on the words to minutes page.
Questions people ask
Is this speech to text free?
Yes, with a fair-use limit of 20 clips per hour per visitor. Each clip can be up to 60 seconds; record several in a row to build a longer note.
How accurate is it?
The transcription model handles clear speech in a quiet room very well, including punctuation. Accuracy drops with background noise, crosstalk or a distant microphone. Always read the result before you use it.
What happens to my recording?
When you press Stop, the clip is sent over HTTPS through Vercel AI Gateway to a speech-recognition model, and only the text comes back. NoteFast does not store the audio or the transcript on a server. The text is kept in your browser like any other note.
Which languages work?
The model detects the spoken language automatically and handles most major languages. Speak one language per clip for the best results.
Does it work on iPhone and Android?
Yes, in current Safari, Chrome, Edge and Firefox. The browser will ask for microphone permission the first time.