Speech recognition, although not a new technology, but the real-time conversion of voice into text dictation transcription has become a new breakthrough point in the vertical scene of artificial intelligence. Recently, Sogou launched the transcription and shorthand "artifact" - Sogou dictation, from the "evolution" of the Sogou input method to the Sogou dictation, the AI ​​application gradually "fly into the home of ordinary people", and the natural interaction also led the AI ​​scene.
When the Sogou input method was officially launched in 2006, users were in the golden age of keyboard input; in 2011, Sogou began to look forward to laying out its own voice technology and quickly productized within one year. From the keyboard to the touch screen, to the voice input, Sogou input method has accumulated experience in the "human-computer interaction" mode, and typing with the mouth has gradually changed from fashion to user habits.
Speech is the most natural way of human communication and human-computer interaction. It is also considered to be the starting point for the era of artificial intelligence. As one of the most powerful AI companies in China, Sogou has established a strong voice self-research team with the largest voice data on the Internet. Statistics show that the daily frequency of Sogou input method has reached 260 million times, an increase of more than 80% from a year ago. Through the large-scale accumulation of high-quality voice training data and deep learning technology, Sogou has also transformed the technical advantages of this speech recognition into more applicable scenarios.
From a technical point of view, the key to Sogou dictation products is the accuracy of speech recognition. It is understood that Sogou dictation uses the long-term speech transfer technology of Sogou input method. From the project to the present, the error rate has dropped by 30%. In the acoustic model, the end-to-end deep neural network technology DeepLC-CLDNN+CTC technology is adopted, and the transfer mode uses the method of DeepCNN+CTC. The language model is modeled based on the T-level massive input method text data using neural network.
Sogou dictation recognition accuracy has reached the international leading level, voice input is faster, more convenient and more accurate than keyboard typing. However, the application process of AI is not completely technology-oriented, but rather a scenario-driven product orientation. The focus is on how to deepen user needs and how to combine scenes more. Only when requirements and scenarios are combined can they become good AI products. In the field of voice, Sogou first realized that the product landing needs scene-driven, in the vertical scene, AI can be really used by users.
In specific application scenarios, Sogou dictation is optimized for different environments used by users, such as meetings, writing novels, etc., and the recognition effect is improved by more than 15%; for libraries, cafes, etc., it is not convenient to speak loudly. The use of voice scenes, providing whisper recognition technology, can still be accurately identified when the person's speaking volume is as low as 30 decibels. Sogou dictation as a multi-scene voice dictation tool, greatly improving user productivity.
From the speech recognition ability of Sogou input method to Sogou dictation, the curtain of natural interaction changing life gradually opened. In the future, voice technology has a lot of opportunities in various application scenarios. For example, in smart home scenes, we hope to use voice and TV, remote control, speakers, curtains, etc. to go home. Not only is the smart home application scenario, but in more vertical application scenarios, such as in-vehicle, medical, education and other environments, the human-computer interaction changes brought by voice will profoundly change our lifestyle and habits.
The ultimate depiction of artificial intelligence by human beings is always the same as human language, which is the development goal of Sogou artificial intelligence. For Sogou input method, AI also gives it more future. In the concept of Sogou, the machine can better understand people's intentions when using input method, so as to push related information, derivative content, and future, Sogou input method. The auxiliary dialogue will help humans communicate better in the machine age.
From input method to Sogou dictation to auxiliary dialogue, Sogou has extended the natural interaction of people through AI technology, which has improved the convenience and timeliness of the device, broadened the practical scene and increased the interactive latitude. What Sogou has been doing is Helping users to “express and obtain information is simplerâ€, focusing on the development of artificial intelligence technology in the field of language, and natural interaction leads the AI ​​application.
Editor in charge: null
Basil Oil,Chinese Herbal Oil,Evening Primrose Oil,Chinese Medicine Oil
Ji'An ZhongXiang Natural Plants Co.,Ltd. , https://www.jxzxessentialoils.com