UNISOUND (09678) Shanhai Zhiyin 2.0 is officially released, reshaping a new paradigm of human-machine interaction.

With the surge of the era of intelligent agents, UNISOUND (09678) is accelerating the completion of its ability puzzle for the "one base, two wings" technology strategy. Following the upgrade of the "Shanhai Zhiyi" 5.0 medical model last year, the company recently released the heavyweight "Shanhai Zhiyin" large model 2.0. It is reported that the "Shanhai Zhiyin" large model 2.0 relies on the multimodal, cross-language base capability of "Shanhai Atlas", focusing on the evolution of three core abilities - understanding professional and regional accents, expressing affection and warmth, and responding with extreme agility. In terms of "understanding", this model's ASR capability has demonstrated leading speech recognition capabilities in both public test sets and its own comprehensive test sets, achieving a leading level from general to extremely comprehensive in the assessments, surpassing mainstream open source and closed source speech large models in China and reaching the highest level in the industry. On the "expression" level, Shanhai Zhiyin-TTS focuses on "highly personified + creative diversity" as its core, currently supporting 12 dialects (including Cantonese, Sichuan dialect, and Shanghai dialect) + 10 foreign languages, and can even switch between 12 Mandarin styles. More importantly, Shanhai Zhiyin 2.0, based on an end-to-end interactive brain, has overcome the challenge of fluent bidirectional interaction, supporting interruptions at any time, immediate responses, and coherent follow-up questions, making human-machine conversations flow like a conversation between close friends. Behind all these capabilities is UNISOUND's unique "Shanhai Atlas" intelligent computing base, which deeply integrates the generic multimodal large model base with the Atlas infrastructure, serving as the foundation for professional intelligent agents and the core of perception AI.