Hong Kong’s HKChat assistant has added a music-generation feature powered by HKGAI’s YuE2 model, according to an October 30 report from Unwire. Users can hum a melody or upload a photograph, add a written description and generate a song with accompaniment and vocals. The feature emphasizes Cantonese music and was reported as free to use at launch.

The two creation routes offer different starting points. Humming supplies a musical idea for the system to develop, while a photograph supplies a scene or mood that can be supplemented with text. This places music generation inside an existing assistant rather than requiring users to begin with a separate production application.

YuE2 predates this integration. Its researchers describe a system that connects explicit musical planning with full-song audio generation, producing a representation of melody and harmony before rendering audio. The current news concerns HKChat’s use of that technology, rather than the model’s first release.

The integration shows how regional-language music tools can reach users through familiar consumer services. The launch reporting does not establish a commercial licensing framework for every generated song or disclose how training-data rightsholders are compensated.