Proton Mail has no Dictate button of its own. Voice typing works through whatever speech engine your browser or operating system provides, and that text is typed into the composer before Proton encrypts anything. That is the detail most guides skip: end-to-end encryption protects the message after you press send, not the audio you spoke to write it. This guide covers how to dictate in every version of Proton Mail, which speech engines process audio on your device versus in the cloud, and a simple workflow for writing sensitive replies by voice without undoing the reason you chose Proton.
How to voice-type in each version of Proton Mail
Proton Mail web in Chrome or Edge on Windows: click into the message body, press Windows + H to start voice typing.
Proton Mail desktop app (Windows / Mac): same system shortcut — Windows + H, or press the Fn / Globe key twice on macOS.
Mac in Safari or the desktop app: turn on Dictation in System Settings, Keyboard, then press the shortcut inside the composer.
iPhone and iPad app: tap the body, then the microphone on the keyboard.
Android app: tap the body, then the microphone on Gboard or your keyboard of choice.
Mail clients via Proton Bridge (Outlook, Thunderbird, Apple Mail): use the client's or the operating system's dictation — Bridge just syncs the result.
Where your voice goes: on-device vs cloud speech
Apple devices with on-device dictation (recent iPhone, iPad, and Apple silicon Macs) process most languages locally — audio does not leave the device.
Windows voice typing (Windows + H) uses Microsoft online speech services; the audio is sent to Microsoft.
Gboard can run offline speech if you download the language pack under Voice typing, Offline speech recognition.
Browser speech in Chrome sends audio to Google servers to transcribe it.
Proton itself never receives the audio — only the finished text you send.
A safe workflow for sensitive replies
Split the email into two passes. Dictate the structure and the non-sensitive wording — the greeting, the context, the ask, the sign-off — at normal speed. Then type the parts that actually need protection by hand: account numbers, medical details, legal names of third parties, passwords, addresses. It takes an extra thirty seconds and keeps the speech engine from ever hearing the sensitive half. If you are on an Apple device with on-device dictation or Gboard with offline speech enabled, you can dictate everything, because the audio stays local.
Dictating long emails that still read well
Say punctuation out loud — "comma", "period", "new paragraph" — rather than relying on pauses. Speak one paragraph per idea and stop between them so you can glance at what landed. For long replies, draft in the free Zahvox tool first, clean up the wording, then paste into Proton: you get a quieter editing space, and the composer only ever receives finished text. Read once before sending; the most common dictation errors in email are homophones (their / there, to / too) and names, which spell-check will not catch.
When dictation does not work in Proton Mail
Nothing appears: the caret is not in the body field. Click into it first, then start dictation.
Browser microphone blocked: allow the microphone for mail.proton.me in the address-bar permissions panel.
Windows + H shows "offline": online speech recognition is disabled in Privacy settings, Speech.
Mac shortcut does nothing: Dictation is off, or the shortcut conflicts with another app.
Text appears in the wrong language: change the dictation language in system or keyboard settings, not in Proton.
