"The computer can be taught to recognize emotions according to the nuances in the voice"

Illustration: pixabay.
Illustration: pixabay.

"The computer can be taught to recognize emotions according to the nuances in the voice," said Dr Pink silver rose, research associate at the Center for Media and Information bThe open university As part of a joint meeting of the Department of Information Systems bTel Aviv-Yafo Academic College וBureau of Information Technology who dealt with voice interfaces.

The event was moderated by Dr Ruthie Gafni, Head of the Information Systems Management Program at the School of Management and Economics of the Tel Aviv-Yafo Academic College.

According to Zilber Rod, "when it comes to visual interfaces, the computer knows how to recognize faces and emotions at a much higher level than in voice interfaces."

"I study the intonation and rhythm of speech - that is, everything that is not the words - not the content, but the how, the music of the speech," explains Dr. Zilber-Vorod.

"It turns out that the same person speaks differently when he is in other positions, for example when I talk to my daughter one vocalization comes out, and with my husband another vocalization. The vocalization will change depending on who I am talking to. The purpose of the research is to teach the computer to listen to nuances in speech."

Dr. Ward Zilber and Rod, research associate at the Center for Media and Information at the Open University. Photo: Moti Sadovsky.
Dr. Ward Zilber and Rod, research associate at the Center for Media and Information at the Open University. Photo: Moti Sadovsky.

"The dominant technology today is speaker recognition," says Zilber-Vorud. "People are sitting in a yeshiva talking, and a computer automatically transcribes their words. For such a transcription to make sense, the computer needs to recognize the speaker and what his role is. For example, in news broadcasts, the speaker can be a presenter, a reporter or a guest being interviewed, and the computer needs to recognize this by the vocal tone. The need is Teach the computer to listen to the nuances in speech so that it recognizes not only who the speakers are, but what the speakers' relationship is to each other."

Later, Dr. Zilber-Vorod described an experiment in which she gave four women who knew each other the role of caregiver and patient, with each of them serving in both roles alternately (the first cared for the second, the second the third, the third the fourth and the fourth the first). The conversations were analyzed by an established algorithm artificial intelligence and reached a recognition level of 71%. Another study simulated a guide trying to explain to his trainees how to navigate According to a map that he can't show him but describe. Here too there were high recognition rates.

Dr Nava Shaked from the Faculty of Engineering at the Holon Institute of Technology, spoke about the need for the computer to use a multitasking interface (Multimodality), meaning that the computer will know how to communicate with people transparently through different input methods; speech, image recognition, and will also know how to connect the input in the different technologies.

Dr. Nava Shaked, Faculty of Engineering at the Holon Institute of Technology. Photo: Moti Sadovsky.
Dr. Nava Shaked, Faculty of Engineering at the Holon Institute of Technology. Photo: Moti Sadovsky.

"We want to allow the user to conduct the dialogue in the way that best suits him. There are many technologies that, when combined together, will provide insight while taking into account the context. We have a triangle that makes use of the technologies that exist in every computer - input technologies, sensors and data that artificial intelligence software processes. If we do it right and produce a whole that is greater than the sum of its parts."

As an example, Shaked said, "Autistic children communicate with robots more easily than with humans. The challenge is how to integrate everything together in real time."

"I researched the whole topic of multimodality in particular in the context of virtual personal aids and avatars. The avatars are a representation of an entity in the virtual world that does something for me. It can represent me and can be my personal assistant but it will always have human-like qualities, and even the feeling is that I have a representation of real interaction. If I'm talking to an avatar who is an expert in the field of medicine, I want him to have the knowledge."

Later, Dr. Shaked presented three projects of her students: an avatar for diabetics that reminds the patient to take pills and measure the sugar level; an application for pregnant women based on an avatar resembling a pregnant woman that helps the woman in the various stages of pregnancy; and, an application that allows you to perform the walking route task in a stress test.

Dr Yoram Lebanon, Chief Scientist of Beyond Verbal He presented the system he developed that allows personality types to be identified based on a person's voice. Later, the system he developed succeeded in identifying menstruating women, men suffering from prostate cancer, dyslexics, and even autistics.

Dr. Yoram Lebanon, Chief Scientist at Beyond Verbal. Photo: Moti Sadovsky.
Dr. Yoram Lebanon, Chief Scientist at Beyond Verbal. Photo: Moti Sadovsky.

In the example of the autistics, "in the study we conducted together with Weizmann Institute, it turned out that there was a shift in the sound spectrum. The spectrum, which is supposed to express love, moved to the world of anxiety and later the system also made an initial identification of Parkinson's and locating people who need catheterization due to the narrowing of the blood vessels in the heart by decoding a 90-second file of their voice."

One of the possibilities, he said, "is to use the system in areas where doctors are lacking, and even when there are, their medical knowledge doubles every 73 days and there is no way that the doctors will be able to follow the developments." Now, he explained, "We have developed a UI that will allow, in exchange for economic agreements, other companies to exploit the system. I count at least 300 medical applications for it, and there is no way we can develop more isolated ones."

Dr Guy Feigenblatt From laboratories IBM (IBM) in Haifa described the developments in the laboratory in the field of applying artificial intelligence to improve business results. "IBM's vision connects to cognitive computing and develops cognitive services that help our customers build better businesses."

Dr. Guy Feigenblatt, IBM Laboratories in Haifa. Photo: Moti Sadovsky.
Dr. Guy Feigenblatt, IBM Laboratories in Haifa. Photo: Moti Sadovsky.

According to him, "The main leap in computer cognition in the last 15 years is that the computer is a tool that will help us in making decisions. It's not that in 15 years the computer will be able to replace humans, but I believe that the computer will be able to help us in making the right decisions."

"Computers are able to learn from structured data and free text, and recognize natural language from text, speech and images and interact with the user."

"A cognitive computer will know how to help us improve and specialize in certain areas and help us in making personal decisions or in organizations," said Dr. Feigenblatt in conclusion.

Leave a Reply

Email will not be published. Required fields are marked *

This site uses Akismet to filter spam comments. More details about how the information from your response will be processed.