Redes neurais auto-organizáveis na visualização da fala como recurso fonoaudiológico
Carregando...
Data
2019-07-10
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Resumo
Speech is an important element of socialization and development of the individual. However, millions of people do not develop it partially or completely. Verbal communication disorders usually appear as secondary anomalies due to other pathologies, such as hearing impairment. Although there is an expressive number of hearing impaired, the studies and use of computational resources in training and speech therapy for speech acquisition and improvement are still timid. In this context, the present work proposes the use of self-organizing neural networks as an aid in the speech-language training process of individuals with speech disorders, through visual feedback. The processes required for the development of the proposed system are described, as well as experiments with Kohonen's Self-Organizing Map, using Mel Frequency Cepstral Coefficients as descriptors of speech characteristics. Some configuration scenarios were defined for both the self-organizing map parameters and the cepstral coefficients, where the topological phonetic map of reduced scope was obtained, which produced a lower percentage of error and misconceptions, being 35.83% and 35.42%, respectively.
Descrição
Palavras-chave
Coeficientes cepstrais de frequência em escala Mel, Mapa auto-organizáveis, MFCC, Reconhecimento automático de fala, SOM, Visualização da fala, Automatic speech recognition, Mel frequency cepstral coefficients, MFCC, Self-organizing maps
Citação
GOMES, Ronaldo Silva. Redes neurais auto-organizáveis na visualização da fala como
recurso fonoaudiológico. 2019. 53 f. Trabalho de Conclusão de Curso (Graduação) - Escola de Engenharia Elétrica, Mecânica e de Computação, Universidade Federal de Goiás, Goiânia, 2019.