Voice

Speak and be heard

Chalie supports full voice interaction — you can speak your requests and Chalie responds with synthesised speech. This makes it feel much more like talking to a person than typing at a text box.

Voice is available through the web interface for desktop and mobile use.

Enabling voice

Open Settings and enable voice. Chalie will download the on-device voice models in the background — this happens once and requires no account or internet connection after that.

Once enabled, a microphone button appears in the input bar.

Speaking to Chalie (push-to-talk)

Press and hold the microphone button in the input bar to record your message, then release to send it. Chalie transcribes your speech and responds as normal.

• “What’s on my calendar tomorrow?”
• “Set a timer for 20 minutes.”
• “Read me my latest emails.”

Hearing Chalie’s responses

Chalie speaks its responses out loud using text-to-speech. For short answers Chalie responds directly. For longer content like email summaries or document contents, it reads a condensed version.

The voice and speed are not currently adjustable through the interface.

Voice models

Chalie uses on-device models for both speech recognition and synthesis — your voice data never leaves your machine.

• Speech-to-text: Moonshine (Whisper-family), running locally
• Text-to-speech: Kokoro v1.0, running locally

The models are downloaded automatically when you enable voice in Settings.

Accessibility

Voice mode makes Chalie accessible to people who prefer not to type or who have difficulty using a keyboard. Chalie aims to handle natural speech patterns, filler words, and corrections gracefully — you can say “actually, I meant…” and it adjusts.