Voice dictation for Gemini Notebook
Double-tap Space and hold, say the question, release: the text is in Gemini Notebook's chat box, unsent, ready to edit. Notebook OS adds three voice tools to the notebook you already use — dictation into the chat box, Term Drill for spoken term practice, and Voice Brain Dump for recorded memos.
Dictation into the chat box
- Open a notebook. A mic pill sits next to the chat box.
- Double-tap Space and hold — or click the pill — and speak. Words appear as you talk: grey while provisional, black once recognised.
- Release to stop. The text stays in the box, unsent; fix names and numbers, then press send yourself.
Where the audio goes
Chrome's Web Speech API sends the audio to Google's speech-recognition service and returns text. Notebook OS receives only the text; it never receives, stores or forwards the audio.
Nine languages
- English (US)
- English (UK)
- Tiếng Việt
- 中文
- 日本語
- 한국어
- Français
- Deutsch
- Español
Pick one in the dropdown next to the mic pill; it is remembered between sessions.
Free plan: 20 dictations a month. Pro: unlimited.
Term Drill: hear the term, say the equivalent
One chat question lists the key terms in your sources. The browser reads them aloud one at a time in a voice you pick, pauses — three seconds by default, adjustable — for you to say the equivalent, then shows it. It speaks; it never turns the microphone on.
- Edit the list, shuffle, loop, export .csv
- Replaying a saved drill is always free
- Free plan: 3 generated term lists a month. Pro: unlimited.
- Useful for interpreters, language learners and exam revision
Voice Brain Dump: a spoken memo as a source
From the side panel, record a memo of up to five minutes. Notebook OS uploads the audio file to your notebook as a source, the way you would drag in a recording, and Gemini Notebook reads it with the rest of your sources.
Free plan: 10 recordings a month. Pro: no monthly limit.
Which voice tool does what
| Tool | Listens to you? | What you get | Where |
|---|---|---|---|
| Voice Dictation | Yes, while you hold the keys | Text in the chat box, not sent | Next to the chat box |
| Term Drill | No — it reads aloud | A spoken drill of your terms, exportable as .csv | Studio Recipes → Term Drill |
| Voice Brain Dump | Yes, while recording | An audio source in your notebook | Side panel |
In development
Voice in every prompt field
⚙ In developmentThe dictation pill follows the focused field — Customize dialogs, source notes, Skill editors — not only the chat box. Same hotkey, same languages.
Meeting & lecture capture
⚙ In developmentRecord a meeting or a lecture from the side panel and land it in a notebook as an audio source with a first-pass minutes recipe queued behind it.
Questions
- Where does my voice go?
- To Chrome and then to Google's speech-recognition service, through the browser's Web Speech API — the same path a dictation button on any web page uses. Notebook OS receives only the recognised text; it never receives, stores or forwards the audio.
- Do I have to grant microphone permission every time?
- No. Chrome asks once for notebook.google.com and remembers the answer. If you dismissed the prompt by accident, clear the site's microphone permission in Chrome's site settings and start a dictation again.
- Why does it not send the message for me?
- Because speech recognition gets names, numbers and technical terms wrong often enough that you want a look before it runs. The text lands in the chat box and stays there; you fix what needs fixing and press send yourself.
- Which languages can it recognise?
- Nine: English (US and UK), Tiếng Việt, 中文, 日本語, 한국어, Français, Deutsch and Español. The choice sits in a dropdown next to the mic pill and is remembered between sessions.
- Does Term Drill record me?
- No. Term Drill uses the browser's speech synthesis to read your terms aloud and never turns the microphone on. You say the equivalent to yourself.
Works with Edge, Brave & all Chromium browsers