On July 29th, Spirent [voice epoch] new technology conference, the successful closing of 3W coffee in Shenzhen. From in-field activities to out-of-field experience, from technology release to experience sharing, the crowd was swaying, and three hours were packed. More than 500 people attended the event to participate in this new technology conference. The conference was hosted by Spirent CEO Gao Shixing, and the technology was presented by Prof. Yu Kai, co-founder and chief scientist of Spirit. Dr. Yu Kai is a Ph.D. student at the University of Cambridge. He is responsible for the US military speech recognition project and the end-to-end human-machine dialogue system technology. He has won many international awards and is also the joint laboratory of intelligent human-computer interaction of Spirent-Shanghai Jiaotong University. principal. Spirent CEO Gao Shixing Yu Kai, co-founder and chief scientist of Yu Bi Chi With the advent of the era of intelligent hardware, the concept of "human-computer interaction" has also been heated up. In the era of mobile internet, due to the richness of interactive modalities and interactive targets, users' demands for the freedom and richness of human-computer interaction are increasing. They are not satisfied with simple search functions, but are gradually changing to "task processing". . But at present, intelligent voice technology faces enormous challenges. A truly free and stable human-computer interaction system not only needs to solve the "perception/expression" of speech, that is, traditional speech recognition and synthesis, but also the intelligence of cognitive understanding and abstract thinking. This requires two stages, one is the optimization of traditional speech technology, making it more humane; the second is the use of dialogue technology, using cognitive interaction intelligence to improve the level of task completion and user experience. So what did the Spirit new technology conference bring to the industry? Far-field and anti-noise technology breakthrough: In the international common noise standard test library, Spirent's latest structured anti-noise speech recognition technology has set an international record and achieved the best results in the world single system. The optimization breakthrough of the model algorithm makes Spirent achieve the effect achieved by using the voice noise reduction chip only by using the software solution, which greatly improves the recognition rate and reduces the cost. More advanced echo cancellation and microphone array algorithm combination: Advanced echo cancellation algorithm combined with microphone array algorithm can effectively track user position, enhance voice signal quality, and achieve high performance universal speech recognition. A variety of solutions combining hardware and software give more choices and possibilities. With software and single-microphone conditions, with big data integration and intelligent algorithms, Spirent can still achieve an extremely excellent experience, with recognition speed and accuracy at the world's leading level. Statistical-based parametric speech synthesis method: Traditional speech synthesis uses a splicing technique to record a large-scale speech library. After slicing, the segments are assembled at the time of synthesis, which often requires a large amount of standard speech. Spirit uses the latest statistical-based parametric speech synthesis method, which can achieve large-scale compression of the model scale, greatly improve the speech coherence, and be able to freely train personalized speech. Multi-round interaction and semantic analysis according to context: In the real scene, the traditional speech recognition plus natural language processing scheme does not achieve a good "human-computer interaction". Based on the leading traditional speech recognition synthesis technology, Spirit introduces interactive cognitive intelligence, which is designed and optimized for the final task completion and user experience through situational understanding and multiple rounds of interaction. Yu Kai introduced the new technology to the participants Yu Kai live demonstration "voice input board" In addition, at this event, Feng Li, a senior analyst of Analysys think tank, unveiled the "Analysis of Intelligent Hardware User Portraits", and Ou Ruibo CEO Wang Xionghui, Yingci Technology CEO Wang Xiaobin, Wei Shite CEO Wen Zheng came to the scene to share Their product experience. During the period, Spirit also announced to the scene a large-scale event to be launched, “Chuangxing Projectâ€, which was jointly sponsored by Spirit, FACE++, Micro Motion, FlexibleCloud, 72, and the Middle Ages. Small developers grow and push for a startup incubator program that is booming in the smart hardware space. Related activities, can focus on the 72-variable intelligent hardware recommendation platform. It is reported that the event will be officially launched in the middle and late August. Analysys senior analyst Feng Li Ou Ruibo CEO Wang Xionghui Ying Xiao Technology CEO Wang Xiaobin Wei Shite CEO Wen Zheng Si Bi Chi Information Technology Co., Ltd. is the top intelligent voice solution provider in China. Cognitive dialogue technology is the development trend of voice interaction, and intelligent hardware is the explosion point of voice interaction. The release of this new technology not only brings a revolution to the industry, but also a redefinition of the so-called "intelligence" of today's "smart hardware"! Shaoxing Tianlong Tin Materials Co.,Ltd. , https://www.tianlongspray.com
December 26, 2020