Real-time, both ways
Bidirectional audio over a WebSocket to the Gemini Live API. Barge in whenever you like; the tutor stops and listens.
Parlo is a real-time voice tutor for your pocket. Earbuds in, phone away, and you're in a conversation with a patient native speaker — in any of 137 languages and 850+ regional accents.


Parlo is designed for the moment your phone is in your pocket. Everything after Start is voice — or an earbud button.
Bluetooth or wired, it doesn’t matter. Parlo routes audio to whatever you’re wearing and follows you when it changes.
A foreground service keeps the conversation alive with the screen off. Play/pause on your earbuds mutes and resumes the mic.
Gemini Live listens and answers in real time. Interrupt it mid-sentence, ask for a repeat, switch languages — all by voice.
Bidirectional audio over a WebSocket to the Gemini Live API. Barge in whenever you like; the tutor stops and listens.
Castilian or Rioplatense, Carioca or Porto, Québécois or Marseillais. Pick the accent you actually want to hear.
Tiny phrases, spoken slowly, with English scaffolding after each one. For the languages where nothing lands yet.
Struggle with a word and the tutor quietly notes it. After the walk, the transcript is mined for anything you missed.
Level, correction style, and 30 Gemini voices — with tap-to-hear previews in your target language.
Your own Gemini key, encrypted on-device. No backend, no sign-up, no one in between you and the model.
Textbook accents are nobody's accent. Parlo's catalogue covers every major regional variety of every language it speaks, with native names and a search that understands places — type “porto”, “Québec” or “Bavaria” and jump straight there. Anything missing can be typed in free-form.
Say “let's switch to Portuguese” mid-walk and the tutor changes language on the spot.





Two quiet mechanisms do it for you. During the walk, the tutor has a silent tool it calls whenever it explains, translates or corrects something. Afterwards, the whole transcript is mined for anything else you stumbled on. Both land in a Suggested tray — keep or skip with a tap.
Every Gemini prebuilt voice, grouped and labelled by character. Tap any of them in Settings to hear a greeting in your target language before you commit.

Parlo has no backend and no accounts. You bring a free Gemini API key; it's stored with Android Keystore-backed encryption and only ever travels to Google's API. Sessions, transcripts and vocabulary live in a Room database on the device.
Parlo is a personal project, built in the open. Clone it, build it, and put it on your phone in a few minutes.
git clone https://github.com/jonaylor89/Parlo.git
cd Parlo && ./gradlew installDebug