China launches Tibetan-language AI system DeepZang

(ECNS) - China has unveiled DeepZang, the world's first Tibetan-language large language artificial intelligence (AI) model, as it steps up efforts to develop AI tools for minority languages.

The model was launched on March 15 in Lhasa, capital of the Xizang Autonomous Region, according to its developer CHOKNOR.

DeepZang supports intelligent interaction in Tibetan, Chinese and English. It integrates multiple functions including AI-powered dialogue, real-time translation, and speech-to-text transcription, said its founder Tenzin Norbu.

The model has been trained on approximately 70 million Tibetan parallel text entries, along with more than 30,500 hours of speech data covering major Tibetan dialects, founder Tenzin said.

"DeepZang has officially passed national algorithm and model filings, becoming the country's first native Tibetan large model to complete both procedures," Tenzin said.

"This means ethnic languages now have their own voice in the AI era," he added.

At the launch event, the World Record Certification Agency (WRCA) awarded the project a certificate recognizing it as the "world's first Tibetan-language large model," organizers said.

Several companies, including regional units of China Mobile and PICC Property and Casualty, have signed cooperation agreements to promote applications of the model, the company said.

(By Zhang Jiahao)