Speak naturally — hear it in the other language
Speaker 1
Speaker 2
Pick a language and voice for each speaker, press start, and just talk.
How it works
- Pick each speaker's language, plus the voice that speaks their translations in the other speaker's language — one side per person.
- Press start and talk naturally. When you pause, or press Done — translate, your words are translated instantly on your device.
- Your words are spoken aloud in your chosen voice, in the other speaker's language, then the microphone automatically switches to them, so the conversation flows back and forth on a single device.
Dictation uses the same Web Speech engine as our text-to-speech generator, translation runs on-device through your browser's built-in AI translator, and playback uses your installed device voices — no servers involved.
Languages & voices
The language list is built from the voices installed on your device, and translation coverage comes from your browser's downloadable language packs — most Chrome and Edge installs cover all major world languages. The first use of a new language pair downloads its pack once; after that it is reused, even offline.
On supported languages every speaker can also choose a Natural AI voice — ten studio-quality voices (Aria, Luna, Leo, Kai, and more) generated entirely in your browser by the same engine as our text-to-speech generator. The first use downloads the voice model once; after that replies take just a few seconds to generate.
Want more languages or better-sounding device voices? Follow the add more voices guide for Windows, macOS, iOS, and Android.
Privacy first
- Translation is on-device: your browser's built-in translation models run locally; your words are never sent to our servers — we don't have any.
- Playback is local: translations are spoken with device voices via the Web Speech API, or with Natural AI voices generated in your browser (WebGPU/WASM) — fully local after the one-time model download.
- Voice input: speech recognition is provided by your browser; some browsers may process audio in their speech service.
- Nothing is stored: the transcript lives only on this page until you clear it or leave.
- Open source: audit the code on GitLab.
FAQ
Is the voice translator free?
Yes. Like the rest of Unlimited TTS it is completely free with no account, no app install, and no usage limit.
Which browsers does the voice translator work in?
It needs a browser with both the Web Speech API and the built-in on-device Translator API — currently Chrome and Microsoft Edge version 138 or newer on desktop and Android. A microphone is required for voice input.
Does it really switch between speakers automatically?
Yes. After your words are translated and spoken aloud, the microphone switches to the other speaker's language automatically, so two people can hold a natural back-and-forth conversation on one device.
Is my conversation private?
Translation runs on-device and playback uses local voices — neither ever reaches our servers. Voice input uses your browser's speech recognition, which in some browsers may process audio in the cloud. The transcript lives only on this page.
Can I use the Natural AI voices in conversations?
Yes. For supported languages, each speaker's voice dropdown includes ten Natural AI voices generated entirely in your browser. The first use downloads the voice model once (progress is shown below the start button); after that each reply takes a few seconds to generate. Device voices remain the instant option.
Which languages can it translate?
The language list comes from the voices installed on your device, and translation coverage comes from your browser's downloadable language packs. Most Chrome and Edge installs cover all major world languages; the first use of a new language pair downloads its pack once and it is then reused offline.