Unisound

中文摘要 ↓

Registered name

云知声智能科技股份有限公司

Unified Social Credit Code

91110108599663854W registered in Beijing 北京

Stock code

• HKEX:09678 Hong Kong Stock Exchange 香港交易所

Website

• www.unisound.com

Weibo

2941129227

WeChat

• 云知声 (Service)
• 云知声Labs (Official · Channel)
• 云知声聪聪智能 (Official)

Models and generation engines · Multi-vertical platform vendors

Unisound (云知声智能科技股份有限公司), founded in 2012 and registered in Beijing’s Haidian district, is a speech-recognition and large-model company listed in Hong Kong. It released U2, an agent-oriented large language model, in June 2026.

Its 2D digital-human generation algorithm, which turns input audio into lip-synced video of a customer-specified likeness for customer service and information broadcasting, passed the Cyberspace Administration of China’s deep-synthesis filing in February 2024. The same month it launched a digital-human solution that clones a presenter’s voice and likeness from a single three-minute recording, and it sells custom 2D, 3D and cartoon digital humans driven by its 山海 model.