OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Voice Agents

Qwen Audio Agent

Qwen Audio Agent is a JavaScript runtime for real-time voice interaction with AI agents. Its voice frontend supports full-duplex, multi-turn conversation, while an optional backend agent can handle delegated tasks and return results to the ongoing conversation. It provides WebUI, terminal TUI, and desktop floating-orb interfaces.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Qwen Audio Agent — OpenViST project profile in Voice Agents.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when building an agent experience where voice conversation should remain active during delegated or longer-running work, or when comparing frontend-only voice interaction with a voice-plus-backend-agent setup.

Who it is for

  • Developers adding real-time voice interaction to an AI agent.
  • Teams prototyping an agent interface that combines speech recognition and speech generation.
  • Agent users who want conversations to continue while backend tasks run.

Documented uses

  • Add full-duplex voice interaction, interruptions, and sustained multi-turn dialogue to an agent.
  • Delegate tool use or longer-running work to a backend agent while maintaining the same conversation.
  • Track task progress, cancel work, and receive completed results in the conversation.
  • Choose a browser, terminal, or desktop interface for interacting with the runtime.
Before you choose

Requirements & limits

The documented CLI quick start requires a DashScope API key for its default voice frontend. The project also describes cloud and self-hosted frontend options with separate setup instructions. The repository is public and identifies the license as Apache-2.0.

  • The voice frontend handles real-time conversation; a separate backend agent handles delegated tasks.
  • Frontend-only operation is listed as an option, while backend settings can be configured separately.
  • The default quick-start voice frontend uses DashScope; other cloud and self-hosted options require their own configuration.
  • The documented quick-start runtime requires Node.js 22.22.2+ or 24.15.0+, and npm 10+.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up a frontend-only voice conversation from the command line

Configure Qwen Audio Agent with the DashScope voice frontend, start its Gateway, and connect a browser or terminal client.

Before you start

  • Node.js 22.22.2+ or 24.15.0+ and npm 10+.
  • A DashScope API key.
  • For terminal audio, follow the TUI platform-specific audio prerequisites.

Steps

  1. Install the CLI with the command below if it is not already installed.

    Documented command
    npm install -g qwen-audio-agent

  2. Run qwenaudio config and open the reported config.env file; its default location is ~/.config/qwaudio/config.env.

  3. Set QWEN_AUDIO_REALTIME_PROVIDER=dashscope, DASHSCOPE_API_KEY=your-key, QWEN_AUDIO_REALTIME_MODEL=qwen-audio-3.0-realtime-plus, and AGENT_PROTOCOL=none in the configuration.

  4. Start the Gateway with qwenaudio and leave it running. In another terminal, launch qwenaudio webui or qwenaudio tui.

  5. In the client, confirm the Gateway and voice frontend are connected, allow microphone access, enable input, and say “Hello.”

Expected outcome

The client should show a transcript and play a voice reply, confirming a frontend-only conversation.

Guide limitations

  • This setup does not start a backend Agent; chat and enabled frontend tools remain available.
  • Each user can have one active client per Gateway; connecting a new client disconnects the previous one.
  • The TUI has platform-specific audio behavior and prerequisites; consult its guide for audio dependencies and mode differences.
Documentation 1 ↗Documentation 2 ↗Documentation 3 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
QwenAudio/qwen-audio-agent · Official homepage title · retained in its source language
How we review sourcesSuggest a correction