In the practical development support project for welfare equipment promoted by the New Energy and Industrial Technology Development Organization (NEDO), IT giant Human Techno System and Tohoku University have introduced a new technology to convert the voice of people with speech disabilities into clear pronunciation. developed.Human Techno System has presented some of its research results at the 116th Electronic Information and Communication Society / Welfare Information Engineering Study Group, and people who have lost their voice due to laryngeal cancer or amyotrophic lateral sclerosis (ALS). It seems to be good news for improving the quality of life.
According to NEDO, the new technology has expanded the function of the voice quality conversion technology developed by Tohoku University, and by converting the voice of a healthy person into the voice of a person with a speech disability, the person with a speech disability can speak with his or her own voice.Until now, a large amount of data was required to generate voice, but even with limited data, high-quality voice generation has become possible.
As a welfare device for people with speech disabilities, there is a personal voice synthesizer that reads out arbitrary text using its own voice data recorded in advance, but the current device has only a function to faithfully reproduce the recorded voice, and pronunciation is unsuccessful. If it is clear, the voice of the synthesizer will also be obscured.
Therefore, Human Techno System and Tohoku University have succeeded in producing clear and high-quality speech by introducing technology to generate speech waveforms from the characteristics of speech and preprocessing of learning data.The human techno system is continuing demonstration experiments at multiple bases, aiming for practical use in 2023.