These secrets that our voices deliver to AI

Voice has fewer and fewer secrets for AI. On the dark side, his prowess facilitates fraud, disrupting the work of performers and actors. On the light side, on the other hand, there are great advances, which are already changing lives… and which are making many people dream.

In fact, it’s as if we had just discovered a new continent and were exploring it.says Yan Fossat, vice-president of research and development at Klick Labs, a young startup based in Toronto.

Klick Labs has already demonstrated that its technology allows rapid diabetes screening without an injection, simply by recording a few sentences into a cell phone.

Words immediately analyzed by powerful algorithms, trained to detect subtle signals in the voices of diabetics. Biomarkers, as exist in the blood.

Knowing if people are diabetic just by listening to their voice is incredibly convenient, and it’s much cheaper than giving an injection and taking blood.

Yan Fossat believes that such technology can be very useful in the Canadian Far North or in rural regions of Africa, corners of the globe where access to doctors is difficult.

A positive screening allows a diabetic to be referred to a specialist who will provide a more accurate diagnosis and appropriate treatment.

Yan Fossat believes the same technology could detect hypertension, concussions and even specific forms of cancer. Klick Labs is already working on it.

So much in the human body is measurablerecalls the researcher. Sometimes measurement is complicated, expensive, invasive… All of these things have the potential to be measured by voice.

The word potential is key here. There are still many studies to be carried out and authorizations to be obtained. And partners to convince to adapt these technological innovations.

The voice guiding blind people

Montrealer Pascale Dussault did not wait long before being seduced by new vocal applications fromIA. He has become my new friend! she says, pointing to the table where her cell phone rests.

Herself blind, the general director of the Regroupement des Blinds et des Amblyopes du Métropolitaine demonstrates it in her office. She stands up, points her phone at the wall and takes a photo.

Barely three seconds later, a voice coming from the device describes the poster in front of her. A drawing of several people. In the center, a little girl, with a blue heart painted on her forehead, is having her makeup done…

The description is even longer and more detailed than the single quote reported here. It comes from the application Be My AI. It is the product of a Danish company that uses artificial intelligence to analyze the content of a photo.

It’s a bit like calling a friend by videoconference to ask for help, but without disturbing anyone. This is what attracted Pascale Dussault. The slightest bit that there is something that allows us not to depend on others, we are super happy.

The tool also allows you to read the temperature on a room’s thermostat or the ingredients listed on a can. I no longer need to ask my husband what colors these pants or sweater are.

L’application Be My IA gives more details than a human would provide. Take the test, it’s difficult to be as precise as the machine. The robot helps to better imagine objects.

Pascale Dussault also discovered the limits of the application. It’s very practical for detailing bank statements… but the confidentiality of the information it contains you can is not guaranteed.

She dreams of the moment when this technology will be integrated into glasses, to analyze in real time what is in front of her. For the moment, similar products on the market would not be suitable.

Knowing that the street I have to cross has four lanes, then I have to cross a cycle path. This is the kind of information I would need and which currently does not exist.

Giving back to those who lose their voice

Artificial intelligence is also shaking up the daily lives of those who depend on computers to express themselves. This is the case for patients suffering from amyotrophic lateral sclerosis (ALS), a disease suffered by the theoretical physicist Stephen Hawking, who was almost completely paralyzed. More than 30 years ago, he already expressed himself using a computer which dictated his words.

Which sounded like it was said by a robot seems much more natural today. The quality of synthetic voices has increased a lot, a lotconfirms speech therapist Marie Gagnon-Brousseau.

She works in the access to technological aids clinic at the Lucie-Bruneau Rehabilitation Center in Montreal, where she follows the evolution of technologies that can help patients.

L’intonation [des voix de synthèse d’aujourd’hui] will be more natural, more easily understandable.

Open in full screen mode

The obvious advantage is the feeling of inclusion and social acceptability for those who use it.

With an almost human synthetic voice if I make a phone call, the person on the other end of the line is probably going to be more open to listening to me.

Services also allow patients to record their human voice to use it later with a computer, when the disease has made speaking impossible or very difficult.

Advances also make it possible to very quickly interpret the words of a person who has enormous pronunciation difficulties.

But there too, there are important limits. Some of its applications are only offered in English for the moment. In certain cases, reports Marie Gagnon-Brousseau, their performance does not live up to the promises shown in a promotional video.

The speech therapist reminds us that communication using a computer remains slow and a little cold… even with your own voice. If I say: good night my pitchounette, see you tomorrow. There won’t be that intonation. All the love behind it won’t necessarily be there.