Open-Qwen-Music Demo

Listening examples from an independent research reproduction of Qwen-Music

GitHub · Model Weights · Dataset · Reference Paper

This Space is maintained by the Open-Qwen-Music reproduction project. It is not an official Qwen demo and is not affiliated with, endorsed by, or maintained by the Qwen team or Alibaba Cloud.

This static gallery presents audio produced by the checkpoints released with this project:

• 22 reconstructions generated from ground-truth semantic tokens;
• 12 end-to-end text-to-music generations using the default sampling settings;
• the corresponding descriptions, tags, lyrics, and durations.

All model outputs displayed here were generated by this project. They are not official Qwen-Music samples and should not be interpreted as reproducing the results reported in the Qwen-Music Technical Report.

Run Locally

The complete inference implementation and prompt format are available in the GitHub repository. Follow its quick start to generate audio with the released weights.

The Space itself has no build step. To preview the static gallery, serve the repository root with any static HTTP server.

Method Reference

This project follows methods described in the Qwen-Music Technical Report. The paper and its reported results belong to the original authors; the checkpoints and samples shown here were produced independently by this project.