Separación de consonantes a través de parámetros acústicos de energía y sonoridad
Date
2012-05-18
Authors
Llanes Machado, René Antonio
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Central “Marta Abreu” de Las Villas
Abstract
Los trastornos motores del lenguaje (TML) son afecciones que producen deficiencia en la
producción de la voz. Las disartrias constituyen un grupo principal entre ellos. Debido a que
estos trastornos influyen negativamente en la incorporación de los pacientes a la sociedad se le
brinda una especial atención a las técnicas que permiten un mejor diagnóstico, seguimiento y/o
rehabilitación de las personas con afectaciones en su comunicación oral.
Los autores Darley, Aronson y Brown (DAB) llevaron a cabo una serie de estudios en las
Clínicas Mayo donde analizaron las características del habla de grupos de pacientes disártricos
desde el punto de vista auditivo. A partir de estos estudios elaboraron una metodología de
diagnóstico a través de valoraciones subjetivas de determinados rasgos perceptuales. Entre los
resultados más notables se puede mencionar que el rasgo de Consonantes Imprecisas (CI),
presentaba la más alta incidencia en 5 de las 7 disartrias analizadas.
Debido a lo anterior, nuestro trabajo pretende mediante técnicas de Procesamiento Digital de
Señales, lograr una medida objetiva de imprecisión articulatoria, altamente correlacionada con la
valoración del rasgo CI. Se ha observado que la imprecisión se manifiesta generalmente a través
sustituciones de las consonantes oclusivas sordas “P”, “T”, y “K” por sus diferentes variaciones
en fricación, nasalización y/o sonorización de las tres consonantes. En la caracterización de las
consonantes se han empleado índices de sonoridad (ISS) y energía relativa (IES) y varios
clasificadores matemáticos para lograr definir regiones de agrupación de las mencionadas
consonantes en un plano de Energía vs Sonoridad arribándose a buenos resultados de separación.
Este resultado constituye un punto de avance en la determinación objetiva del rasgo CI lo cual
puede constituir una herramienta para el diagnóstico y supervisión de pacientes disártricos.
The motor disorders of language (TML) are conditions that cause deficiency in the production of voice. The dysarthria are a major group among them. Because these disorders have a negative effect on the incorporation of patients to society. A special attention is given to techniques for improved diagnosis, monitoring and / or rehabilitation of people with this affectations in oral communication. The authors Darley, Aronson and Brown (DAB) conducted a series of studies at the Mayo Clinics where they analyzed the speech characteristics of dysarthric patient groups from the standpoint of hearing. From these studies developed a diagnostic methodology through subjective assessments of certain perceptual features. Among the most notable may be mentioned that the trait of imprecise consonants (CI), showed a higher incidence in 5 of the 7 analyzed dysarthria. Because of this, our work aims, through techniques of Digital Signal Processing, to achieve an objective measure of articulatory imprecision, highly correlated with the CI trait assessment. It has been observed that the inaccuracy is usually manifested by substitutions of voiceless consonants "P", "T" and "K" by different variations in frication, nasalization and / or voiced of the three consonants. In the characterization of the consonants were used indices of sonority (ISS) and relative energy (IES) and several mathematical classifiers to achieve defined regions of the aforementioned group of consonants in a plane of energy vs. sonority. Finally obtaining a good separation results. This result is a breakthrough point in the objective determination of CI feature which can be a tool for diagnosis and monitoring of dysarthric patients.
The motor disorders of language (TML) are conditions that cause deficiency in the production of voice. The dysarthria are a major group among them. Because these disorders have a negative effect on the incorporation of patients to society. A special attention is given to techniques for improved diagnosis, monitoring and / or rehabilitation of people with this affectations in oral communication. The authors Darley, Aronson and Brown (DAB) conducted a series of studies at the Mayo Clinics where they analyzed the speech characteristics of dysarthric patient groups from the standpoint of hearing. From these studies developed a diagnostic methodology through subjective assessments of certain perceptual features. Among the most notable may be mentioned that the trait of imprecise consonants (CI), showed a higher incidence in 5 of the 7 analyzed dysarthria. Because of this, our work aims, through techniques of Digital Signal Processing, to achieve an objective measure of articulatory imprecision, highly correlated with the CI trait assessment. It has been observed that the inaccuracy is usually manifested by substitutions of voiceless consonants "P", "T" and "K" by different variations in frication, nasalization and / or voiced of the three consonants. In the characterization of the consonants were used indices of sonority (ISS) and relative energy (IES) and several mathematical classifiers to achieve defined regions of the aforementioned group of consonants in a plane of energy vs. sonority. Finally obtaining a good separation results. This result is a breakthrough point in the objective determination of CI feature which can be a tool for diagnosis and monitoring of dysarthric patients.
Description
Keywords
Trastornos Motores del Lenguaje (TML), Disartrias, Consonantes Imprecisas (CI), Parámetros Acústicos, Energía, Sonoridad, Herramienta de Diagnóstico