André Luis Maciel Leme1, Márcio Abud Marcelino2,3, Pedro Paulo Leite do Prado3. 1. Instituto Federal de São Paulo - IFSP - Bragança Paulista (SP), Brasil. 2. Universidade Estadual Paulista - UNESP - Guaratinguetá (SP), Brasil. 3. Universidade de Taubaté - UNITAL - Taubaté (SP), Brasil.
Abstract
PURPOSE: This study presents the margins of minimum and maximum tolerances for the frequencies of the first three formants (F1, F2, and F3) in the pronunciation of vowels of Brazilian Portuguese for use in voice therapy for the deaf. METHODS: The frequencies were obtained from the voluntary collaboration of 53 adults who had their voices recorded and converted into digital signals during the phonation of each of the seven vowels (/a/, /e/, /Ɛ/, /i/, /o/, /ᴐ/, /u/) sustained for approximately one second. The samples were divided into two groups: male and female. The recording and extraction of the formants were conducted by software developed exclusively for this purpose in MATLAB platform using the eight-coefficient LPC (Linear Predictive Coding) algorithm. RESULTS: The results showed that a consistent reference for the mean values of the F1, F2, and F3 frequencies can be obtained through graphical and statistical analysis of the samples collected from the voice signals. CONCLUSION: The reference values obtained were analyzed and can be used for the calibration of devices and serve as a basis for training oralization for the deaf.
PURPOSE: This study presents the margins of minimum and maximum tolerances for the frequencies of the first three formants (F1, F2, and F3) in the pronunciation of vowels of Brazilian Portuguese for use in voice therapy for the deaf. METHODS: The frequencies were obtained from the voluntary collaboration of 53 adults who had their voices recorded and converted into digital signals during the phonation of each of the seven vowels (/a/, /e/, /Ɛ/, /i/, /o/, /ᴐ/, /u/) sustained for approximately one second. The samples were divided into two groups: male and female. The recording and extraction of the formants were conducted by software developed exclusively for this purpose in MATLAB platform using the eight-coefficient LPC (Linear Predictive Coding) algorithm. RESULTS: The results showed that a consistent reference for the mean values of the F1, F2, and F3 frequencies can be obtained through graphical and statistical analysis of the samples collected from the voice signals. CONCLUSION: The reference values obtained were analyzed and can be used for the calibration of devices and serve as a basis for training oralization for the deaf.