Voice Notes

About Voice Notes

Voice Notes is a free, browser-based tool that listens to you speak and builds a running, numbered list of one-line summaries alongside the full transcript. It records your microphone in short ~16-second segments, transcribes each one with on-device Whisper speech-to-text, then condenses it into a single sentence — using a built-in extractive summarizer by default, or a connected local text model for nicer phrasing. Everything (recording, transcription, and summarizing) runs locally in your browser with nothing uploaded and no signup, and audio is discarded segment-by-segment so even an hour-long session stays light on memory. Use it for meetings, lectures, brainstorms, or quick voice memos where you want the gist captured as you talk.

How to use Voice Notes

1. Enable on-device transcription when prompted — the tool downloads the Whisper speech-to-text model the first time so it can run locally in your browser.
2. Click 'Start listening' and allow microphone access in your browser's site permissions.
3. Talk normally; a pulsing 'Listening…' indicator and elapsed timer show it's recording, and every ~16 seconds a new one-line summary is added under 'Summary notes'.
4. Watch the status (Transcribing… / Summarizing…) and the queued count if segments are still being processed.
5. Expand 'Full transcript' at the bottom to read the complete word-for-word text.
6. Click 'Stop' to end the session and flush the final segment.
7. Use 'Copy notes' to copy the numbered summaries, or 'Download .txt' to save them as voice-notes.txt.

Frequently asked questions

Is my audio uploaded to a server?

No. Recording, transcription, and summarizing all happen on your device in the browser — your audio is never uploaded, and each audio segment is discarded right after it's transcribed.

Is it free and do I need an account?

Yes, it's completely free with no signup, account, or watermark.

Do I have to download anything to use it?

The first time you turn on transcription, the tool downloads the on-device Whisper speech-to-text model so it can run locally. After that it's cached and works offline.

How does it keep a long recording light on memory?

It records in short ~16-second segments and processes them one at a time, dropping each audio blob after it's transcribed — so only the tiny text accumulates, not hours of audio or PCM.

How are the summaries generated, and can I make them better?

By default each segment is condensed to one sentence using a built-in extractive summarizer. If you connect a local text model (the 'generate' capability), it's used to rewrite each segment into a cleaner one-sentence summary instead.

Can I get the exact words, not just the summary?

Yes. Below the summary notes there's a collapsible 'Full transcript' section showing the complete word-for-word text plus a word count.

People also search for

Voice Notes is also known as voice notes to text, speech to text meeting notes, voice note to meeting notes, voice memo converter, voice notes ai app, transcribe and summarize meetings, lecture notes from recording.