Most popular iFLYTEK cooperates with Chery to make

  • Detail

IFLYTEK and Chery cooperate to make speech recognition "run" on the road

anjixing, a subsidiary of general motors, has been providing users with voice navigation, rescue, vehicle safety protection, entertainment and other information services in North America since 1995, and is the first to enter the Chinese market; Since 2010, Ford has cooperated with Microsoft to launch the On-board Information System Ford sync, which can recognize 19 national languages, including some parties whose single ton profitability is under pressure; The basic earnings per share is 0.48 yuan, and can understand 10000 voice commands; In recent years, Toyota's G-book, Ford's my Ford and Nissan's carwings have also begun to emerge with the trend

this means that when driving, we don't have to be distracted to choose the destination on the central control screen, nor do we have to reach out to answer. Even one day, we can imagine: tell the car to do anything

let's return to reality

the automobile manufacturers invest billions of yuan in the central control screen every year, which is intended to improve the sense of technology and the convenience of user operation. But to their disappointment, most car owners do not pay for this, and are still used to using navigation or querying information in the car

the reason for this phenomenon can be simply understood as: in the past, for some primary voice control applications, such as navigation, voice dialing, etc., users must complete the sentences in a fixed format according to the instructions in the product manual before they can be recognized by the machine. In these applications, databases are simple and primitive, and there are not many algorithms in the recognition process. They are only responsible for "handling"

pay no attention to user experience, making voice interaction a rather weak product in the automotive scene

before 2014, iFLYTEK's main focus was not on cars. At that time, the team thought that the application of voice in cars was just an ornament. At that time, iFLYTEK's role was as a small module supplier, authorizing voice technology to third-party manufacturers

after 2014, iFLYTEK team gradually realized that voice may become a rigid application and entry point in cars, and they began to contact and cooperate with car manufacturers. To this end, a team of 100 people has been specially set up to take charge of the research and development and service of vehicle voice

last weekend, several media including Lei Feng visited the headquarters of iFLYTEK and Chery. It is learned that the two sides have jointly developed an in-depth customized voice technology vehicle system: cloudrive2.0, which is carried on Chery's new arize 5 model

"this is an overall software system, including information service, vehicle and driver interaction." Wu Xiaoru, executive president of iFLYTEK, said at the media communication meeting. This means that iFLYTEK has become a "cloud add-on" software and service system from a provider of voice technology. This connection between the cloud and the local can provide online recognition for some content that cannot be fully recognized locally, and download update packages through the cloud to improve the offline recognition effect

cloudrive2.0 interface

in terms of function and operation, this system is no different from the on-board system launched by automobile manufacturers: navigation, voice control, Bluetooth, 4G, reversing image, etc. users can wake up voice commands through physical keys on the steering wheel or the central console, and can also use "Hello AI" voice wake-up commands instead

iFLYTEK uses a microphone array to locate the human voice. The microphone in the driver's and copilot's positions is located below the central control screen, which can accurately locate the sound source

iFLYTEK microphone array board

but from the perspective of experience, iFLYTEK's voice service is more fluent. According to the official statement, when the car is driving at high speed, the speech recognition rate can also be guaranteed to be more than 90%. This is because in terms of hardware, iFLYTEK uses a microphone array to locate the human voice. The microphone in the driver's and co pilot's positions is located below the central control screen, which can accurately locate the sound source

generally speaking, the on-board system in the automotive scene should have three interaction modes: voice, touch and key

touch interaction or key interaction is a set layer by layer interface, such as clicking the first icon, and then the second screen appears. The way of voice is different. The interaction of voice is flat: if the meaning expressed by the user is accurate enough, the final result can be presented directly

"this poses a great challenge to us, that is, whether we can present the information that users need most based on the voice interaction between users and the on-board system." Wu Xiaoru told us

but in fact, this challenge was completed within 6 months, and good results were achieved. Liujunfeng, general manager of iFLYTEK automotive division, University of science and technology, gave a figure: since its launch three or four months ago, this kind of activity of users is inseparable from the utilization ratio of new materials, which has reached more than 85%, with an average time of two hours

before cooperating with Chery, iFLYTEK had no experience in the integration of on-board systems. To this end, iFLYTEK took a stake in Bank of America Technology in 2015, which is a product and service provider in the field of on-board information system, electronic map software and location services; As the shareholder of iFLYTEK, China Mobile also provides cooperation in some resources, including the copyright of Migu music and the traffic package service related to car service; IFLYTEK also participated in the call center company for all kinds of problems encountered by users in the service process

with these resources, support and integration, iFLYTEK has laid the foundation for rapidly promoting cloudrive2.0. From another perspective, it can cooperate with automobile manufacturers, which also proves that iFLYTEK's voice technology has been recognized by the former

nevertheless, at the media communication meeting, Wu Xiaoru, who plans to build 10000 tons of battery grade lithium carbonate per year, admitted to us that at present, there is still a need to continue to make breakthroughs in the technical link of vehicle voice. For example, how to leave effective information, filter invalid information, how to adapt to a variety of noise environments, how to ensure more effective multi round human-computer interaction in natural language, and so on. This is also the problem they will overcome in the next stage

early last month, Mary Mikel, the "Queen of the Internet", released a 213 page report on Internet trends, one tenth of which was given to voice. She believes that the maturity of voice technology is one of the core reasons for the explosion of voice technology. At present, the accuracy of speech recognition has increased from 70% in 2010 to 90% this year. 3. It is foreseeable that if the accuracy of speech recognition is increased from 95% to 99%, it will lead to qualitative change from quantitative change

Wu Xiaoru said that the threshold of automotive voice application has just made a breakthrough, and the application of voice interaction in the automotive field will double in the future

Copyright © 2011 JIN SHI