Dialog mode

Dialog mode lets you speak to your model and hear replies. Speech recognition runs on your iPhone or iPad; the recognized text is sent to your LM Studio server.

Start a spoken conversation

1. Open a chat with a ready model and a working server connection.
2. Use the voice-chat control in the composer to start Dialog Mode.
3. Allow microphone and speech-recognition access when prompted. On-device recognition must be available for the device and recognition language.
4. Speak when the app is listening. It transcribes your turn, sends the text, waits for the answer, and speaks the reply when text-to-speech is enabled.
5. Use Stop voice chat to end the session and return to normal text entry.

Choose how replies sound

Open Settings → App → Voice Chat:

• Dialog Mode TTS turns spoken assistant replies on or off.
• Assistant Voice chooses from installed Apple text-to-speech voices. Preview a voice in the picker before selecting it.

Turning off TTS changes playback, not the server connection or the text stored in your chat. Speech input still needs microphone and speech-recognition permissions.

Privacy and availability

LLMster requires on-device speech recognition rather than falling back to a remote transcription service. The recognized text travels to your configured server just like a typed message. Speech playback uses the selected system voice.

If listening will not start

Check the app’s microphone and speech permissions in system Settings. If the app reports that on-device recognition is unavailable, use text input; changing the LM Studio model will not add speech-recognition support to the device.

Keep the app active while using Dialog Mode. It is an interactive conversation, not an always-listening background service. See Features and data for further checks.