Chat and voice chat

Talk to a local AI model, by voice or by text

Chat has Voice and Text tabs. Start speaking, pick a persona, and every turn is kept in the record panel.

Direct answer

EdgeSpeak chat runs on a model you choose. Voice chat listens, replies and speaks back. Text chat works the same way from the keyboard.

Personas for different conversations

Choose a preset such as Buddy, Voice companion, Emotional support, Language tutor or Roleplay, or write your own prompt. It applies from the next reply.

Live interpretation

Switch the persona to Interpreter, choose a language pair and a style, and speak.

You choose the models

Set the chat model, the transcription model and the voice in the chat settings. When all three are local, the conversation stays on your computer; a cloud model uses its own service.

Keep a record

The record panel shows each turn as it happens.

FAQ

Does voice chat need the internet?

Not when the chat, transcription and voice models are all local. If you connect a cloud model, that part uses its service.

What is Buddy?

A preset persona for friendly, everyday conversation. You can edit it or write your own.

Can I use it as a language tutor?

Yes. There is a Language tutor preset, and the interpreter persona handles two-way interpretation.

Can my own tools connect?

Yes. The local gateway provides a Realtime WebSocket API.

Evidence links