Shanghai University has developed a new version of a large language model called Xiao Hu, which is dedicated to the preservation of the Shanghai dialect, the university said on March 10 during a demonstration session held on its campus.

By incorporating the Shanghai dialect into its development, the model is able to process multi-language text conversion and features continuous improvements in the recognition, reading, and generation of the Shanghai dialect, according to Zeng Jun, director of Shanghai University's publicity department, who introduced the technical background and future applications of Xiao Hu at the event.

Zeng said that in the future, Xiao Hu will be able to be used as a voice assistant, broadcaster, or a teaching platform. He also highlighted its potential in the research front, such as generating content in the Shanghai dialect and as a resource for Haipai or Shanghai-style culture.

In the demonstration session, Xiao Hu showcased its ability to recognize and generate the Shanghai dialect.

Federico Demarco, an alumnus of the university from Argentina who has experienced the large model on three separate occasions, said that it is a good tool for learning the Shanghai dialect, as there are few resources online he could refer to.

Duan Yong, deputy Party secretary of Shanghai University, said that the launch of Xiao Hu, which merges AI technology with the Shanghai dialect, contributes to the city’s digital revolution while also helping preserve Shanghai’s cultural heritage.

Moving forward, the university said it will further strengthen the model by making the pronunciation more accurate, improving emotional expression, and expanding use cases to include opera, poems, and the recognition of various versions of Shanghai dialect.