Tencent Sogou Input Method officially announced full AI transformation, and released the 20.0.0 commemorative version. This upgrade focuses on three key areas: AI voice, AI translation, and AI typing, leveraging the capabilities of Tencent Hunyuan large model to achieve a significant leap in input efficiency and intelligent experience.

image.png

In the new version, AI voice input stands out especially. Based on the AI voice large model developed by Tencent Hunyuan, the recognition delay has been reduced by 40%, and the overall accuracy has been increased to 98%. The accuracy of dialect recognition has also improved by 30%. Notably, the new version has been specifically optimized for scenarios like "library" and "late night" where soft spoken words are needed: even at as low as 20 decibels, the recognition accuracy can still remain stable at 97%. In addition, the model can intelligently reorganize disordered speech and provide one-click replacement functions for easily confused sounds such as “de”, “di”, and “de”, making expressions more professional.

Regarding AI translation, Sogou Input Method deeply integrates the champion-level translation model of Hunyuan, supporting over 30 languages. The translation function is seamlessly embedded into the input process, allowing users to achieve "input and translate" without switching apps while chatting or reading, greatly improving cross-language communication efficiency.

In the field of AI typing, through a deep understanding of context and user intent, the candidate word results have become more accurate. The preference rate of popular words has increased by more than 10%, and it also realizes daily updates of movie titles, celebrities, and internet buzzwords, truly achieving "today's new show, tomorrow's typing".

Aside from the comprehensive breakthroughs in AI capabilities, this version also introduced an experience optimization with great sincerity - one-click closure of keyboard ads, further returning to the pure essence of the input tool.