AI will retain the voices of people who have lost the ability to speak

The generated speech is associated with the "computer" voice of the late Stephen Hawking. However, soon people who have lost the ability to speak will be able to maintain their own voice. Artificial intelligence will help them in this.

Computer-generated speech will relieve the dumbness of patients whose voice apparatus is damaged due to injuries or illnesses. Now, unfortunately, the new voice has nothing to do with the former, which makes the patient and his family feel uncomfortable. However, the situation will soon change.

According to the Guardian, the first of its kind clinic opened in the United States, the main purpose of which is to preserve the voices of patients with the help of artificial intelligence. Its creation was the fruit of the joint efforts of Northeastern University in Boston and VocaliD.

Professor Rupal Patel, the founder of VocaliD and the main researcher of the clinic, notes that the company already allows people to record their voices at home before they lose their ability to speak.

Unfortunately, many customers do not have the right equipment. Others turn to the company too late, and they do not have enough time to keep their voice. A new clinic should solve these problems and make technology more accessible.

The approach that will be used in the center is much more complicated than slicing and subsequent combining of audio fragments. At the first stage, patients will have to record their own voice lasting 2-3 hours. For example, it can be a recitation of poems or stories. Based on these notes, specialists will create a machine learning algorithm that will generate speech similar to the patient’s voice. He will be able to pronounce any words, including those that are not on the original recording. The client will simply upload a digital voice to a smartphone or other device and voice phrases typed using it.

According to Patel, the technology is improving rapidly. Digital voices increasingly resemble the original. Researchers are currently working on filters to simulate intonation. In addition, they would like the voices to mature and age with the wearer. While the team has not learned how to turn the voice of a child into the voice of a teenager or adult, however, it is already possible to convert it into the voice of a slightly grown child.

The voice recording service will be free, but the creation of a digital voice will cost $ 1499. During the first year, customers will not be charged. Anyone can save their voice in the clinic, but specialists will give priority to those who really risk losing their ability to talk in the near future.

nanonews.net