Conversation Mode

Conversation Mode lets you have a continuous, hands-free voice conversation with NoteTalker. You speak, NoteTalker listens, thinks, responds out loud, then automatically listens again. No buttons to press between turns — just talk naturally.

Note

Conversation Mode runs entirely on your device. Speech recognition, the language model, and speech output all run locally — nothing you say is sent to a server.

How It Works

1. Click the conversation mode button (the circular arrows icon) in the chat panel
2. Start talking — NoteTalker begins recording immediately
3. Pause when you’re done — NoteTalker detects silence and stops recording automatically
4. Your speech is transcribed on your own device and sent as a chat message
5. NoteTalker processes your message with the local language model and speaks the response aloud
6. When the response finishes playing, NoteTalker automatically starts listening again
7. The loop continues until you stop it

This creates a natural back-and-forth rhythm, similar to a phone call or in-person conversation.

Starting and Stopping

• Start: Click the conversation mode button (circular arrows) — it turns blue when active
• Stop: Click the button again, or NoteTalker will exit automatically after a period of extended silence

You can also type messages while conversation mode is active. NoteTalker will respond to typed messages normally and continue the voice loop afterward.

When to Use Conversation Mode

Conversation mode is ideal when you want to:

• Think out loud — brainstorm, process ideas, or talk through a problem
• Stay hands-free — cook, exercise, drive, or work with your hands while talking to NoteTalker
• Have a longer discussion — explore a topic over multiple exchanges without clicking buttons each time
• Dictate and refine — speak your thoughts, then iterate on them conversationally

Tips for a Smooth Experience

• Use headphones or earbuds to prevent NoteTalker’s spoken response from being picked up by your microphone
• Speak clearly and pause when you’re finished — NoteTalker uses silence detection to know when you’re done
• Keep background noise low for better transcription accuracy
• Check your silence settings in Settings > AI & Voice if NoteTalker cuts you off too soon or waits too long — you can adjust the silence threshold and duration

How It Differs from Push-to-Talk

Feature	Push-to-Talk	Conversation Mode
Recording	Manual start/stop	Automatic (silence detection)
Response	Text only (unless TTS is on)	Always spoken aloud
Next turn	You click the mic again	Automatic — starts listening after response
Best for	Quick one-off questions	Extended back-and-forth discussions

Requirements

• A working microphone
• The voice models downloaded on first launch — Whisper Base for speech-to-text and Kokoro for text-to-speech — alongside your local language model
• Enough GPU VRAM (or Apple Silicon unified memory) and system RAM to run the language model, speech-to-text, and text-to-speech at the same time

Everything runs on your own hardware. Once the models are downloaded, Conversation Mode works fully offline — you can turn off networking and the back-and-forth keeps going.