Publication: Procesamiento de voz para mejorar la pronunciación
Loading...
Official URL
Full text at PDC
Publication Date
2017
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
El aprendizaje de la pronunciación de idiomas extranjeros mediante programas informáticos se encuentra en pleno desarrollo en la actualidad. Sin embargo, se trata de un campo que no se encuentra muy desarrollado en el software libre. La mayoría de aplicaciones de aprendizaje de idiomas que ofrecen esta utilidad tienen software propietario.
Por esta razón, en este proyecto haremos un esfuerzo por investigar y llevar a cabo una de las vías para conseguir esta utilidad. La aproximación que le daremos a este problema será la comparación de audios con las voces de un nativo y un estudiante. Se tratarán las dos voces para facilitar su comparación, esta comparación nos servirá a los estudiantes para saber si nuestra pronunciación se aproxima a la de un nativo y podernos enfocar en qué aspectos mejorar.
Para ello, veremos cómo hemos de tratar señales de voz y qué tipo de filtros usar para limpiarlas. Después aplicaremos la transformada de Fourier y construiremos espectrogramas con el objetivo de introducir el tratamiento de imágenes para la simplificación de la información y al final poder hacer la comparación de los datos con la información destacada. Este proceso se implementará como una aplicación Java.
Learning of foreign language pronuntiation through computer programs is in full development nowadays. Nevertheless, it is a field not very developed as free software. Most of the existing language learning programs which offer this utiliry are propietary software. For this reason, in this project we will make an effort to investigate and accomplish one of the ways to achieve this utility. The approach we will give to this problem will be the comparison of the audio given by a native speaker with that of a student. Both voices will be treated to facilitate the comparison, a comparison which will help the students to know if their pronuntiation is close to that of a native speaker, and will let them focus on which aspects to improve. For this, we will see how we must treat voice signals and what types of filters to use in order to clean them. Next, we will apply Fourier transform and build spectrograms with the purpose of introducing image treatment for information simplification, to get to do the comparison between the data and the highlighted information. This process will be implemented as a Java application.
Learning of foreign language pronuntiation through computer programs is in full development nowadays. Nevertheless, it is a field not very developed as free software. Most of the existing language learning programs which offer this utiliry are propietary software. For this reason, in this project we will make an effort to investigate and accomplish one of the ways to achieve this utility. The approach we will give to this problem will be the comparison of the audio given by a native speaker with that of a student. Both voices will be treated to facilitate the comparison, a comparison which will help the students to know if their pronuntiation is close to that of a native speaker, and will let them focus on which aspects to improve. For this, we will see how we must treat voice signals and what types of filters to use in order to clean them. Next, we will apply Fourier transform and build spectrograms with the purpose of introducing image treatment for information simplification, to get to do the comparison between the data and the highlighted information. This process will be implemented as a Java application.
Description
Trabajo de Fin de Grado en Ingeniería Informática y Matemáticas (Universidad Complutense, Facultad de Informática, curso 2016/2017)