XJTU's Muxing SYKI-SPEECH team is dedicated in voice AIGC area and has developed the SYKI-SVC singing voice conversion technology.

The 2025 World Artificial Intelligence Conference recently held in Beijing saw fierce competition among 280 teams and over 500 humanoid robots from 16 countries. At the opening ceremony live on China Central Television (CCTV) on Aug 14, the "Xunxiao" humanoid robot, developed by the Robotics Institute of Xi'an Jiaotong University (XJTU), presented a visual feast that blended cutting-edge technology and artistic aesthetics.

Xunxiao emerged from YouiBot, a company founded in 2017 by XJTU alumnus Zhang Zhaohui. Relying on the company's strength in scene orientation and technological breakthroughs, YouiBot has established partnerships with more than 300 companies, covering all production links in subdivided fields such as substrates, epitaxy, masks, and packaging and testing of the first, second, and third generations of advanced electronic manufacturing.

YouiBot has also jointly established the Embodied Intelligence Robot Research Institute with XJTU, and released a plan for seven humanoid robots at this year's China Embodied Intelligence Conference, with Xunxiao debuting as the first product.

Xunxiao can complete diversified tasks such as cleaning and security, troubleshooting, dust-free handling, and material scheduling, promoting the intelligent upgrading of industrial and household scenarios and marking a new era in multimodal task execution by embodied robots.

On the same day, the "2025 China • AI Grand Ceremony" was broadcast on CCTV-1. During the ceremony, the performance of the classic song "Nessun Dorma" driven by AI technology attracted high attention.

Its core technology is the SYKI-SVC singing voice conversion technology proposed by the Faculty of Electronic and Information Engineering of XJTU at ICASSP 2025. This technology can replace the timbre with the timbre of a specified singer while retaining the skills of the original singer, and achieve ultra-high sound quality and skill restoration through original high-frequency post-processing technology, making the singing voice generated by AI almost indistinguishable from a real person.