Second, says Tomashenko, researchers are studying distributed and federated learning, where your data doesn’t come out of your device, but machine learning models still learn to recognize speech by sharing their training with a larger system. . Another approach is to build an encrypted infrastructure to protect people’s voices from espionage. However, most efforts focus on voice anonymization.
Anonymization tries to make your voice sound human while removing as much of the information that could be used to identify you as possible. Currently, speech anonymization efforts involve two separate lines: anonymizing the content of what someone is saying by deleting or replacing any sensitive words in the files before saving them, and anonymizing the voice itself. Most voice anonymization efforts right now involve passing someone’s voice through experimental software that will change some of the parameters of the voice signal to make it sound different. This may involve altering the pitch, replacing speech segments with information from other voices, and synthesizing the final output.
Does anonymization technology work? Undoubtedly, the male and female voice clips that were anonymized as part of the Voice Privacy Challenge in 2020 sound different. They’re more robotic, sound slightly painful, and could, at least for some listeners, be from a different person than the original voice clips. “I think it can already guarantee a much higher level of protection than doing nothing, which is the current state,” says Vincent, who has been able to reduce the ease of identifying people in anonymization research. However, humans are not the only listeners. Rita Singh, an associate professor at the Institute of Language Technology at Carnegie Mellon University, says that total unidentification of the voice signal is not possible, as machines will always have the potential to establish links between attributes and individuals. all connections that are not clear. to humans. “Is anonymity with respect to a human listener or is it with respect to a machine listener?” says Shri Narayanan, a professor of electrical and computer engineering at the University of Southern California.
“True anonymization is not possible without completely changing the voice,” Singh says. “When you completely change your voice, then it’s not the same voice.” Despite this, it is still worthwhile to develop voice privacy technology, Singh adds, as no privacy or security system is completely secure. Fingerprints and facial recognition systems on iPhones have been falsified in the past, but they are still an effective way to protect people’s privacy.
Goodbye, Alexa
Your voice is increasingly being used as a way to verify your identity. For example, a growing number of banks and other companies are analyzing your voice fingerprints, with your permission, to replace your password. There is also a possibility that voice analysis will detect the disease before other signs are evident. But the technology to clone or falsify someone’s voice is advancing rapidly.
If you have a few minutes of someone’s voice recorded or, in some cases, a few seconds, it is possible to recreate that voice through machine learning:The Simpsons’ voice actors could be replaced by deep false voice clones, for example. And commercial voice recreation tools are available online. “There’s definitely more work to be done in identifying speakers and producing voice-to-text and text-to-speech than in protecting people from any of these technologies,” Turner says.
Many of the voice anonymization techniques that are currently being developed are still a long way from being used in the real world. When they are ready to be used, it is likely that companies will have to implement tools themselves to protect the privacy of their customers; there are currently few individuals who can do to protect their own voice. Avoiding calls with call centers or companies that use voice analysis, and not using voice assistants, could limit the amount of your recorded voice and reduce the chances of an attack.
But the biggest protections can come from legal cases and protections. The European GDPR covers biometric data, including people’s voices, in their privacy protections. The guidelines say that people should be informed of how their data is used and give their consent if they identify themselves, and that some restrictions should be placed on personalization. Meanwhile, in the United States, Illinois courts, which have some of the strongest biometric laws in the country, are increasingly inspecting cases involving people’s voice data. McDonald’s, Amazon and Google are facing legal scrutiny over how they use people’s voice data. Decisions in these cases could set new rules for the protection of people’s voices.