(Yicai) March 27 -- Kunlun Tech, the Chinese internet and gaming company behind the Opera browser, launched the world's first music reasoning model under its Mureka series.

Mureka O1 is the first music model to introduce chain-of-thought technology, which allows it to incorporate thinking and self-criticism in the inference process to enhance music quality, creation efficiency, and flexibility, the Beijing-based company announced yesterday.

MusiCoT, a variant of the CoT method developed by the Mureka team, pre-generates songs' music structure before finalizing the auto tokens for decoding, thus improving the structural coherence and instrumental arrangement of the generated music.

Kunlun also unveiled Mureka V6, which is the base model of Mureka. It incorporates the company's self-developed in-context learning technology and supports pure music generation and AI music creation in 10 languages, including English, Chinese, Japanese, and French.

Mureka O1 and V6 both support a variety of music styles, including jazz, electronic, pop, country, rhythm and blues, and rock. They offer a music reference function and a vocal function. The first allows users to upload audio files or YouTube links as references, and the second allows them to choose from a selection of official singer timbres or upload their own voices for the AI to learn and replicate.

Kunlun will continue to increase its investment in AI model capabilities to maintain its leading position among top-tier music generation models globally, Chairman and Chief Executive Officer Fang Han said.

The competition in the AI music generation field is fierce. Kunlun released the first generation of its Mureka model series, Mureka V1, in April last year. Since then, ByteDance launched mobile audio creation tool Sponge Band and Tencent Music Entertainment Group's Tianqin Lab co-develop a music AI model with Tencent AI Lab.

Editor: Futura Costaglione