VSING (08292) announced that the Group is developing an AI-driven singing evaluation technology, designed to assess vocal performances through a more comprehensive and human-perception-aligned model. Differing from traditional scoring systems that primarily focus on a limited set of technical metrics, the Group's AI evaluation technology adopts a multi-dimensional assessment philosophy, simultaneously encompassing both technical execution and expressiveness.

The Board believes that singing is a complex form of human expression that transcends quantifiable technical accuracy. By utilizing AI for holistic performance evaluation, the Group aims to enhance the understanding and assessment methods of singing quality in real-world environments.

The Group considers the capability to develop this AI technology dependent on a robust and differentiated data foundation. Through its existing platform and physical venue network, the Group has accumulated over 2 million real-world singing-related data records, including historical singing activities and a vast number of user performance recordings.

This large-scale dataset of authentic user performances enables the Group to continuously train and optimize its AI models based on the nuances of live singing and its perception. The Board is of the view that the depth and authenticity of this data are key factors driving the long-term, more effective development and sustained competitive edge of the Group's AI evaluation technology, whereas peer solutions primarily rely on limited or synthetic datasets for development.

This initiative represents a forward-looking development that supports the Group's long-term strategy at the intersection of technology, creativity, and human performance.