Sogou opens dictation service, the voice recorder industry enters the AI ​​era

On August 28, 2019, Sogou held a press conference on the opening of Sogou Dictation Service and AI Innovation Alliance with the theme of "New Voice, AI Future". At the conference, Sogou opened its core technical capability "Sogou Dictation" service and announced the establishment of an AI Innovation Alliance with four leading voice recorder companies, namely Patriot, Newmine, Sony Voice Recorder, and Wancheng Group. The convening of this press conference marks that Sogou is leading the voice recorder industry into the AI ​​era.

[[275215]]

At present, the peak voice request of Sogou Input Method has exceeded 700 million times, ranking first in the world's Chinese voice applications. Relying on this, the transcription technology behind Sogou Dictation Service has taken the lead in the industry. Its real-time voice transcription uses Sogou's self-developed delay-controllable Adaptive Attention-based end-to-end modeling technology, and the non-real-time voice transcription function uses the industry-leading deep Transformer-based end-to-end voice recognition technology, with a voice transcription accuracy of more than 95%; in terms of voice enhancement, Sogou's self-developed Smart Voice microphone array algorithm can perform multiple depth optimizations on noise and reverberation to ensure high-fidelity restoration of human voice. In terms of intelligent auxiliary editing, Sogou voice transcription can provide multiple candidate words for low-confidence transcription results through cloud-based intelligent error correction editing technology, based on word graph search and input error correction model, to help users correct the transcription results. At the same time, it also has good automatic speaker distinction, intelligent text smoothing and punctuation prediction capabilities, which can provide users with more convenient functions.

With its powerful voice technology and excellent software and hardware integration capabilities, Sogou launched the AI ​​voice recorder C1 in March this year. Its transcription accuracy is as high as over 95%, and it has multiple AI functions such as real-time mutual translation, cloud sharing, and simultaneous interpretation of Chinese and English. In a very short period of time, it has achieved very large single product transactions on multiple e-commerce platforms. It can be seen that Sogou's voice transcription technology has been tested and recognized by the market. At this press conference, Sogou launched an upgraded version of the AI ​​voice recorder Colorful Edition to meet users' more diversified and more fashionable needs. At the same time, it opened up the market-tested "Sogou Dictation" service system behind the AI ​​voice recorder, which brought about the AI ​​transformation of the entire voice recorder industry. This is an effective attempt by Sogou to implement its AI strategy.

It is worth mentioning that this time Sogou Dictation Service has achieved the connection with Sogou Input Method, which not only brings huge advantages in traffic and data to the dictation service, but also is an extension of Sogou Input Method's mission of "expressing information more simply", and is also the advantage of Sogou Dictation Service compared to other voice transcription platforms. At present, Sogou Input Method has a 99% market share on the PC side and more than 450 million daily active users on the mobile side. Based on the user usage habits of Sogou Input Method, Sogou Dictation Service can automatically build a personalized recognition network, reducing the error rate of users' common expressions by 40%. After Sogou opens the dictation service, it can provide access services for more than 90% of voice recorder products on the market, allowing all users in need to enjoy the experience of "recording for 1 hour and producing manuscripts in 5 minutes".

In addition, at the press conference, Sogou and several leading companies in the voice recorder industry reached an AI innovation alliance, which is also a major subversion and upgrade for the business model of the voice recorder industry. The alliance will enable the voice recorder industry to transform from a single revenue model based on hardware to a multiple revenue model of hardware + services, promote the voice recorder industry to achieve comprehensive intelligent upgrades, and deeply integrate with AI innovation. Through AI voice recorder innovation, opening Sogou voice transcription services, and establishing an AI innovation alliance, Sogou is leading the voice recorder industry into the AI ​​era from three directions: products, technology, and industrial chain.