In China, officials and companies have deployed facial-recognition technology to a degree uncommon elsewhere, both commercially and—more controversially—for widespread surveillance. Now it’s the turn of voice recognition.
The government of the southwestern Guizhou province, Tsinghua University, and Beijing-based d-Ear Technologies announced they are collaborating on a pilot project that will link unique voice features to people’s national ID information and create, maintain, and secure a database of “voiceprints.”
China’s digitally-connected population has enabled the government to track people in many forms, such as using facial recognition to prevent toilet paper theft and jaywalking. But d-Ear (link in Chinese) and other companies say that voice recognition is a better and cheaper way to verify identity. Cheng Ge, marketing director of d-Ear, founded in 2002, told Quartz this week that the d-Ear pilot would be similar to social media giant Tencent’s December pilot program in the southern city of Guangzhou, which allowed people to use facial recognition within its all-purpose WeChat app to create a digital version of their government-issued ID card. They could then use the digital ID for hotel check-ins and airport security checks, among other things.
While the WeChat pilot took place in Guangzhou, at the forefront of public use of facial recognition, Guizhou will be the pilot province for voice recognition before a possible larger-scale application, said Cheng. The company this month began building a data center to store voice data in Guizhou, a province that is increasingly positioning itself as a tech hub. It’s home, for example, to Apple’s first China data center, where a data storage firm established by the local government will manage local users’ iCloud data.
Other Chinese companies have also been testing out voice identification. In December, Alibaba, China’s largest e-commerce company, announced that it was working to install voice-recognition technology in ticket machines for all metro stations in Shanghai to verify commuters’ identities. One of the biggest players in the field, iFlytek, provides voice signatures for medical information in eastern Anhui—and also provides voice-recognition services for police bureaus in Xinjiang, the region home to the heavily surveiled Uyghur ethnic minority. Outside China, as of 2014, government and corporate databases had stored some 65 million voiceprints, the Associated Press reported that year.
While convenient, these sorts of uses of biometric technology raise concerns about the trade-off with privacy. It seems a natural corollary that any biometric identification technology could eventually be used in surveillance in China, though Cheng says the company is only looking at self-verification uses, for example, social security access, hotel check-in, and train ticket purchases.
Cheng says his company has to work with the government because of the “sensitive nature” of biometric data, and says that the voice data belongs to users—but companies and government entities could have access to it “for specific business purposes.”