Development of a 'AI Speech Recognition-Based Evaluation Program for Children with Speech Disorders'
Oct 19, 2024
A research team led by Professor Jang Dae-hyun of the Department of Rehabilitation Medicine at the Catholic University of Korea's Incheon St. Mary's Hospital has developed an AI (Artificial Intelligence) speech recognition-based evaluation program for children with speech impairment (artificial speech impairment).
Speech disorder is a disorder that has difficulty communicating due to inaccurate speech sounds, and for children with speech disorder, early diagnosis and treatment are important as it can negatively affect social interaction and learning ability if they do not receive proper treatment during language development.
Currently, speech disorders are diagnosed by language rehabilitators who directly 轉寫 (話) children's speech and analyze errors.
Professor Jang Dae-hyun's research team wanted to increase the efficiency and accuracy of articulation tests by incorporating AI voice recognition technology into speech disorders diagnosis with the Korea Research Foundation's excellent new research project and the Ministry of Trade, Industry and Energy's bio-industry technology development project starting in 2022.
The research team secured a total of 90 minutes of voice data from 137 children with speech impairment to develop an AI voice recognition evaluation program and used it to develop and upgrade an end-to-end (E2E) voice recognition model. In addition, to verify the program's performance, an articulation evaluation was conducted on 30 children who visited the rehabilitation medicine department at Catholic University of Korea's Incheon St. Mary's Hospital with articulation disorders. In this process, a study was conducted to compare the direct evaluation of language rehabilitators and the results of evaluation analysis through AI voice recognition models.
As a result of the study, the analysis results of the language rehabilitator and AI voice recognition model showed about 90% agreement, confirming the possibility of articulation evaluation using AI.
This evaluation program is currently in the process of improving user experience and stability after conducting a medical device suitability evaluation conducted by the National Rehabilitation Center.
In addition, Professor Jang Dae-hyun's research team is also developing an AI treatment program as a follow-up study. The AI treatment program is being designed to be easily used at home and is being developed with the aim of supporting systematic home treatment training.
Professor Jang Dae-hyun expected that the AI voice recognition program developed this time will play an important role in improving language development and prognosis of children with speech disorders by quickly and accurately analyzing speech patterns of children with speech disorders.
Meanwhile, the research 'Automatic Speech Recognition for the Diagnostics of Pronunciation in Korean Children' was published as an academic paper in Clinical Linguistics & Phonetics.
Speech disorder is a disorder that has difficulty communicating due to inaccurate speech sounds, and for children with speech disorder, early diagnosis and treatment are important as it can negatively affect social interaction and learning ability if they do not receive proper treatment during language development.
Currently, speech disorders are diagnosed by language rehabilitators who directly 轉寫 (話) children's speech and analyze errors.
Professor Jang Dae-hyun's research team wanted to increase the efficiency and accuracy of articulation tests by incorporating AI voice recognition technology into speech disorders diagnosis with the Korea Research Foundation's excellent new research project and the Ministry of Trade, Industry and Energy's bio-industry technology development project starting in 2022.
The research team secured a total of 90 minutes of voice data from 137 children with speech impairment to develop an AI voice recognition evaluation program and used it to develop and upgrade an end-to-end (E2E) voice recognition model. In addition, to verify the program's performance, an articulation evaluation was conducted on 30 children who visited the rehabilitation medicine department at Catholic University of Korea's Incheon St. Mary's Hospital with articulation disorders. In this process, a study was conducted to compare the direct evaluation of language rehabilitators and the results of evaluation analysis through AI voice recognition models.
As a result of the study, the analysis results of the language rehabilitator and AI voice recognition model showed about 90% agreement, confirming the possibility of articulation evaluation using AI.
This evaluation program is currently in the process of improving user experience and stability after conducting a medical device suitability evaluation conducted by the National Rehabilitation Center.
In addition, Professor Jang Dae-hyun's research team is also developing an AI treatment program as a follow-up study. The AI treatment program is being designed to be easily used at home and is being developed with the aim of supporting systematic home treatment training.
Professor Jang Dae-hyun expected that the AI voice recognition program developed this time will play an important role in improving language development and prognosis of children with speech disorders by quickly and accurately analyzing speech patterns of children with speech disorders.
Meanwhile, the research 'Automatic Speech Recognition for the Diagnostics of Pronunciation in Korean Children' was published as an academic paper in Clinical Linguistics & Phonetics.
|
bellho@sportschosun.com
