Fintech
Corsound can match a voice to a face to fight online scammers
Your face is no longer yours. And not even your voice.
Scammers can steal both from the Internet. And, using the power of Artificial intelligencethey can create a deep false version of you so convincing it gets past bank security.
Banks and financial institutions today are increasingly using voice biometrics to confirm the identity of callers.
And criminals are increasingly exploiting this voice biometric data in two ways.
They clone the caller’s voice to trick the bank’s computer into thinking they are an existing customer.
And they create a completely fake identity to open a new account.
Corsonea startup based in Tel Aviv, has a solution for both problems. First, it developed technology to detect a cloned voice.
Criminals take a voice sample from social media, a YouTube video, or even an online meeting.
They then clone it using free online software to impersonate the real customer.
Corsound can distinguish between a real voice and a cloned one.
Second, it can determine whether a photo of the customer opening a new account matches the person’s voice.
So, if a scammer uses A’s image and B’s voice, they will immediately reject the request.
This is because it knows what a person will sound like based on their appearance.
Create a face from a voice
It’s quite surprising if you translate it into your own life experience. You and I pick up the phone to call a stranger and we have no idea what he looks like. But Corsound does.
And it is currently working on cutting-edge technology that will be able to draw a sketch of someone’s face, based on no more than five or six seconds of voice recording.
A sketch generated from an audio voice file. Photo courtesy of Corsound
It will listen to you on the phone – which is often poor quality audio – and create a picture of what you look like. It’s not matching your voice to an existing photo in a database. It’s generating an image from scratch, instantly.
The company describes it as “magic” and “the only technology in the world that can create a face simply from your voice.”
Orel Agmon Halido, sales manager at Corsound, points out that this is still in development – which is why he was disappointed when I asked for a personal demo – but provides an idea of how it works.
“We’re like a musical instrument,” he says. “Our voice comes through our lungs and throat, and it’s influenced by the shape of our face, our oral cavity, our nasal passages, our lips, all this area. We basically train our model to distinguish between all these differences.”
The AI learns, by processing countless examples, which faces correspond to which voice. And which ones aren’t.
“If you do it multiple times, thousands and tens of thousands of times, you can figure out what a voice looks like,” Halido says.
“Our goal is basically to draw a person’s face, but I’ll be very honest with you. We still need a lot of data to do this.”
Switching to commercial mode
The technology is based on 2019 research from MIT (Massachusetts Institute of Technology) on the correlation between voice and face.
Researchers have found that a short snippet of audio reveals a huge amount of data about a person’s gender, age, ethnicity, skin tone, nose and jaw shape, and more.
Corsound was founded in June 2020, as a subsidiary of the Tel Aviv-based AI company Cortica, to develop ways to implement the technology. It currently has 17 employees in Tel Aviv, most of whom work on research and development based on over 200 AI patents owned by the parent company.
Corsound has so far raised $3.5 million from Canadian venture capital firm Awz Ventures and Israel Innovation Authority.
With some proof of concept (POC) ready, according to Halido, the company is now moving from development to commercial mode and looking for customers.
“It’s the first time, as far as we know, that there’s technology that can prevent identity theft,” he says.
The threats posed by Generative AI they are already huge and are getting bigger and bigger.
“Banks should do everything they can to protect customers. And that’s why they need to embrace new technologies to understand the threats that come with AI technology.”
Synthetic voice alert
While banks have traditionally been quite conservative in adopting technology, he says, banks and financial firms are Corsound’s largest target market.
The company presented its technology to potential banking customers in February at the Finovate 2024 conference in London, with fake calls from customers.
In the first, an Australian woman calls to wire $2,000 to her husband. The rumor is false, but she deceives existing calling technology which, according to Corsound, currently has a 90 percent market share.
The same recording is then played back on Corsound’s technology and immediately flagged as “Fraud: Synthetic Voice Warning” because it knows the voice has been digitally manipulated.
Halido then simulates the onboarding process for a new banking customer using its company’s technology.
“My identity is safe because my voice is my password,” he says to the bank computer. “Please verify me.” Corsound immediately verifies that his voice is authentic, because it is.
He then uploads a “stolen photo” – an image of the company’s CEO Gal Haselkorn – instead of a photo of himself.
The AI immediately rejects your request because it knows the face provided doesn’t match the voice.
According to Halido, Corsound’s technology will save the financial and banking industry billions of dollars a year in fraudulent transactions.
Voice-to-face technology also has broader applications and could revolutionize law enforcement investigations by generating facial structures and facial sketches from voice recordings alone.