iFLYTEKChairman Liu Qingfeng revealed that iFlytek will release iFlytek Spark V4.0 on June 27 this year. This new version will showcase the latest end-to-end intelligent voice technology achievements, and iFlytek also plans to launch its end-to-end voice technology modeling solution in the next three quarters. In addition, iFlytek will also provide users with sentence replication, high-noise scene voice recognition, and multi-dialect and multi-language free switching functions that GPT-4o does not have.
In terms of super-anthropomorphic synthesis technology, Liu Qingfeng proudly stated that iFlytek has reached the international leading level in the fields of Chinese, English and multiple languages. In terms of full-duplex technology, iFlytek defined the human-computer interaction standards in the era of the Internet of Everything as early as 2016, including far-field, multi-person, full-duplex and other features. It is worth mentioning that iFlytek successfully led and proposed the international standard for full-duplex voice interaction in 2020, and was officially issued in May 2023. These advanced voice technologies will be fully demonstrated in iFlytek Spark V4.0.
Looking ahead, Liu Qingfeng revealed that iFlytek will focus on research and development in the fields of far-field high-noise multi-person speaking scenarios, highly expressive personalized scenarios, cloud-edge-end integration, and software-hardware integration. iFlytek will build a comprehensive and sustained leading advantage by relying on its technological advantages, market opportunities, and implementation results. According to him, the iFlytek Spark V3.5Max version launched by iFlytek at the end of May has surpassed the latest version of GPT-4Turbo0429 in terms of text generation, knowledge questions and answers, logical reasoning, and mathematical ability, demonstrating its strong technical strength and market competitiveness.