You Can Modify Your Face But Not Your Voice



The Speech Technology Center, headquartered in St. Petersburg, Russia, was at Milipol this week showing its newest product—multi-modal biometrics that combine voice identification and facial recognition to “ensure the best identification results,” according to Anna Belash, head of communications. 

It’s easy to modify a face: grow facial hair, pluck eyebrows, wear colored contact lenses, modify hair color and hairstyle, “but it’s absolutely impossible to modify your voice,” she told me. Every person’s voice is as unique as their fingerprints, and the Speech Technology Center—which started out 21 years ago doing research on voice and speech—quickly saw how interested law enforcement authorities were in their work.

“We have a number of competitors,” Belash said, “but we are the only ones to have implemented a nationwide voice ID system.” The system, which works even with speech fragments just 6 sec. long and of poor quality, is deployed across Mexico to help law enforcement officials identify and track drug traffickers and kidnappers. “There is huge interest in Latin America for this VoiceGrid ID system,” she said.

The company produces recording and transcription systems (Cesar is a computer transcriber while Nestor is a real-time distributed speech transcription system) as well as noise suppression and speech enhancement software and hardware, which can also be used to clean up old recordings. They made me listen to the the clearest version I've ever heard of Neil Armstrong’s words as the Eagle floated down to the surface of the Moon on July 20, 1969.

Source: AviationWeek