Getting Started

Get Talkative Lobster running on your machine in a few minutes.

Prerequisites

OpenClaw Gateway

Talkative Lobster uses OpenClaw as its LLM gateway. You need to install and run it before using the app.

1. Install OpenClaw by following the instructions in the OpenClaw repository
2. Configure at least one LLM provider (e.g., Anthropic, OpenAI, Ollama) in ~/.openclaw/openclaw.json
3. Start the gateway — it runs on ws://127.0.0.1:18789 by default
4. Note your gateway token from ~/.openclaw/openclaw.json under gateway.auth.token

Auto-loading keys

If OpenClaw is installed, the app can auto-load your GATEWAY_TOKEN and API keys from ~/.openclaw/openclaw.json — just click the OpenClaw button in the Settings modal.

Install

Download and install the app for your platform from the Download page. For detailed OS-specific instructions, see the Installation page.

First Launch

On first launch, the Settings modal opens automatically and walks you through:

1. OpenClaw Gateway — enter your GATEWAY_TOKEN (default URL: ws://127.0.0.1:18789)
2. STT Provider — choose a speech-to-text provider and enter API keys if needed
3. TTS Provider — choose a text-to-speech provider and configure voice settings

Each section has a Test button to verify the connection. All three checks must pass before you can start.

See the Providers page for detailed setup instructions and available voices.

Usage

Conversation Flow

Once setup is complete, the main screen shows a waveform visualizer and a status indicator.

1. Click the mic button (or the green ON button) to enable your microphone
2. The app runs a brief noise calibration (~1.5 seconds) — status shows "Calibrating..."
3. Once ready, status changes to "Listening..." — start speaking naturally
4. When you stop talking, your speech is transcribed (status: "Recognizing...")
5. The LLM processes your message (status: "Thinking...")
6. The AI speaks back through your speakers (status: "Speaking...")
7. After the response finishes, the app returns to "Listening..." and waits for you to speak again

Interrupting the AI

You can interrupt the AI at any time while it is speaking — just start talking. The app will immediately stop the current TTS playback and begin processing your new input.

Stopping

Click the STOP button to cancel the current operation. This works during any active state (recognizing, thinking, or speaking) and returns to the idle state.

Settings

Click the gear icon in the bottom-right corner to reopen the Settings modal at any time. Changes take effect immediately — no restart required.

Next Steps

• Providers — detailed STT and TTS provider configuration
• Configuration — all settings explained
• Troubleshooting — common issues and solutions
• Contributing — development setup for contributors