Quick Start

If you haven't installed yet, see Installation first.

1. Create Configuration

qwenaudio config

The command will display the configuration file path and create a config.env template with comments.

2. Fill in Configuration

The minimal configuration only requires a DashScope API Key:

DASHSCOPE_API_KEY=your-key

When you need to execute backend tasks, select a backend agent and specify the backend model:

DASHSCOPE_API_KEY=your-key
# Voice frontend model: flash for lower latency and cost savings, plus (default) for better quality
QWEN_AUDIO_REALTIME_MODEL=qwen-audio-3.0-realtime-plus
# Backend agent: leave empty or set to none to start in frontend-only mode
AGENT_PROTOCOL=openclaw
# Backend model: explicit values use standard ACP; empty reuses Agent config
QWEN_AUDIO_AGENT_BACKEND_MODEL=qwen3.7-max

The default uses the DashScope real-time voice frontend; you can also switch to a local speech-to-speech frontend, which does not require a cloud API Key.

3. Start

Start the Gateway in one terminal:

qwenaudio

Open another terminal and start the TUI:

qwenaudio tui

You can also use the browser interface (default http://127.0.0.1:3101):

qwenaudio webui

Frontend-Only Mode

When AGENT_PROTOCOL is not set (or set to none), the Gateway only provides real-time voice chat. Requests that require backend execution will return a clear explanation and will not create tasks or guess execution results. You can also explicitly start in frontend-only mode with qwenaudio --backend none.

For selecting, one-click installing, permission modes, and persistent service of backend agents, see Backend Agents. For a complete list of environment variables, see Configuration. For TUI platform differences, see TUI Notes; for the browser client, see WebUI.