Car voice recognition "performance" is not as good as that of a thousand-yuan mobile phone, and technology improvement is difficult

Recently, according to the Automotive News website, although voice recognition technology has been introduced into the automotive field for several years, even now this technology is still plagued by various minor problems.

While Google Now, Apple's Siri and Microsoft's Cortana already provide excellent services for millions of smartphone users, automakers are still on the road to developing reliable and effective voice recognition systems.

Nuance Communications launched a speech recognition technology upgrade package last year specifically for the redesigned BMW 7 Series.

According to the 2016 U.S. Initial Quality Survey released by JD Power, 23% of all complaints from car buyers involved car information systems, and voice recognition system issues also accounted for a considerable share. Renee Stephens, vice president of U.S. car quality survey, said in an interview with reporters: "In our opinion, failures caused by voice recognition systems are still the top complaint issue for car consumers. This year, automakers have made some improvements in this area, but progress is still slow."

Rainey-Stephens also noted that complaints surrounding voice recognition systems aren't just coming from older car owners. In fact, voice recognition was named one of the top five most difficult features to use in cars by people born in the 1960s and 1970s, as well as millennials and baby boomers.

Nuance Communications, based in Burlington, Mass., is the world's No. 1 supplier of voice control systems for cars, and says voice recognition technology is improving all the time. Arnd Weil, general manager of Nuance Automotive, said the voice recognition feature in the BMW 7 Series sedan is an example of how the technology can improve. The company launched a voice recognition upgrade package for the redesigned BMW 7 Series last year.

Arden Weir, General Manager, Nuance Automotive

He introduced that the upgraded voice recognition software can understand the spoken commands issued by the car owner, convert the car owner's voice into short messages and send them out, and the car owner can even temporarily add voice commands during the system execution of commands. Although there is still an imbalance in functional improvements, reliability has improved. For example, the above-mentioned voice system has a very high ability to recognize telephone numbers, and its recognition accuracy for voice dialing is as high as 90% to 95%. For routine commands, today's voice recognition system has enhanced its ability to understand and execute, but it is still a bit awkward to execute those commands that are not very programmed, such as finding some "interesting places" as required by the car owner.

Early voice recognition systems relied on databases to execute voice commands. The data in these databases was relatively simple, such as the geographical locations of chain stores such as Starbucks, Exxon gas stations and Hilton hotels, etc. The new in-car voice recognition system is connected to the cloud server. Under the premise of accurately understanding the voice command, it can find any commercial location in the cloud server.

In fact, there are a series of problems in actual use by users. For example, the foreign names of some restaurants are very difficult to understand, so the voice recognition system is naturally powerless. Another problem is that when the sound in the car is relatively noisy, the function of the voice recognition system will also be affected. For example, when other passengers are talking in the car, the voice commands issued by the owner himself cannot be accurately recognized by the system.

To solve the above problems, car manufacturers began to increase the number of microphones in the car from one to two, just like the two ears of a human being. The voice recognition system uses two microphones to locate different speakers, thereby more accurately identifying the driver's voice. Arden-Vale said that this new voice recognition system is called "beam-forming" and it was first used in Audi, Mercedes- Benz and BMW models. At present, some mass brands in the automotive industry have also begun to use this system.

Voice recognition will be further improved with the use of more powerful computer chips. Because cloud connectivity is not always available, automakers have developed a voice recognition backup system using computer chips and installed it in the car. This system is less sophisticated than the cloud voice recognition system, but its capabilities will continue to improve.

Despite the continuous improvement, the performance of vehicle voice recognition systems is unlikely to be comparable to that of a $200 (about 1,000 yuan) mobile phone, because the noise in the car is inevitable and the driver will definitely pay more attention to the road traffic conditions. Even with the unremitting efforts of companies like Nuance, it will be a slow process for JD Power to improve the rating of vehicle voice systems.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.