Enhancing Speech Recorded from a Wearable Sensor Using a Collection of Autoencoders

González-Salazar, Astryd; Gutiérrez-Muñoz, Michelle; Coto-Jiménez, Marvin

doi:10.1007/978-3-030-41005-6_26

Cited by 3 publications

(4 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Como tercera etapa, se creó una actividad de evaluación, la cual no tiene las características de una evaluación regular de este curso, las cuales constan típicamente de ejercicios de matemáticas o de aplicaciones en esquemas, circuitos u otros sobre los cuales se deben aplicar conceptos matemáticos para obtener una solución cerrada. El planteamiento consta más bien de un conjunto de datos generado como parte de los proyectos, sobre el cual se plantea a los estudiantes la verificación de uno de los resultados generados en una publicación académica reciente [7], así como la reflexión de la importancia de realizar una tarea como ésta con un enfoque social. Finalmente, se planteó un cuestionario de apreciación de la actividad a los estudiantes del curso.…”

Section: Marco Metodológicounclassified

“…Actividad evaluativa. Se realizó una asignación de evaluación a los estudiantes, en la cual se les brindó un conjunto de datos proveniente de señales de habla, los cuales fueron parte de la publicación mencionada [7]. Este conjunto de datos consiste en valores de las señales de habla captada por el dispositivo de comunicación y reconstruida utilizando varios métodos.…”

Section: Actividad Generada Para La Integración De Actividades Sustan...unclassified

See 1 more Smart Citation

Integración de actividades de investigación y acción social en la docencia de Ingeniería Eléctrica

Coto Jimenez

2021

rev.digit.educ.ing

Self Cite

View full text Add to dashboard Cite

El objetivo del presente artículo es mostrar los resultados de una experiencia docente desarrollada en la Escuela de Ingeniería Eléctrica de la Universidad de Costa Rica, diseñada para integrar la investigación y la acción social con los contenidos teóricos de un curso de Matemática Superior. Esta experiencia ha sido posible gracias a la interrelación de proyectos inscritos en la Escuela de Ingeniería Eléctrica, así como a la colaboración multidisciplinaria con áreas como la Educación Especial y la Orientación. La finalidad de la experiencia ha sido propiciar el aprendizaje significativo del estudiantado con la integración de los ejes sustantivos de la universidad: docencia, investigación y acción social. Dada la percepción favorable de los estudiantes y el docente en la actividad, se sugiere abrir espacios de reflexión y análisis en otras áreas de especialidad para llevar a las aulas los resultados de sus proyectos y el trabajo multidisciplinario. De esta manera, se puede exponer al estudiantado la problemática de investigación y acción social, donde los conocimientos impartidos encuentran su aplicación y razón de ser.

show abstract

Section: Marco Metodológicounclassified

Section: Actividad Generada Para La Integración De Actividades Sustan...unclassified

Integración de actividades de investigación y acción social en la docencia de Ingeniería Eléctrica

Coto Jimenez

2021

rev.digit.educ.ing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Considering these key themes and the importance of real-world applications that can be understood and addressed by younger learners and general audiences, we propose the exploration of audio and speech in teaching strategies on AI concepts, especially those based on classification methods. For this purpose, we selected a real-world problem that arose from the need to help people with disabilities, and the opportunity to provide them with alternative and augmentative communication systems [6].…”

Section: Introductionmentioning

confidence: 99%

Exploring the potential of an audio application for teaching AI-based classification methods to a wider audience

Coto-Fernández¹,

Coto-Jiménez²

2022

View full text Add to dashboard Cite

Knowledge about artificial intelligence (AI) is becoming increasingly important for many careers, especially those based in science and engineering. Besides formal education, the impact of AI on society lead to consider educational projects for teaching the fundamental concepts of AI at wider audiences, including high school levels. This can help more general audiences to better understand how AI works, with the hope that also parents and educators can help students develop a healthy appreciation for implications and limitations, along with an appropriate relationship and deeper interest on it. In this paper, we present a pilot project for teaching an AI-based classification method that is empirically evaluated with real data of a real problem, which can be understood and tackled with basic mathematical tools and activities suitable for high school students. With this proposal, we aim to show how audio and speech applications can inform a wider audience about advances in AI, its characteristics, and its future impact on society. Results and lessons learned from this project can form the basis for further projects using different tools and data, according to students’ interests and initiative.

show abstract

“…In this work, a method based on deep neural networks (DNN) was proposed to map the noisy speech to clean speech [4,5]. As mentioned, with degraded signals, speech technologies do not work properly, for this reason, the DNN approach can be implemented for better results in several applications, such as in mobile phone applications, speech recognition systems, and assistive technology [6,7].…”

Section: Introductionmentioning

confidence: 99%

Assessing the effectiveness of transfer learning strategies in BLSTM networks for speech fenoising

Coto-Jiménez¹,

González-Salazar²,

Gutiérrez-Muñoz³

2022

View full text Add to dashboard Cite

Denoising speech signals represent a challenging task due to the increasing number of applications and technologies currently implemented in communication and portable devices. In those applications, challenging environmental conditions such as background noise, reverberation, and other sound artifacts can affect the quality of the signals. As a result, it also impacts the systems for speech recognition, speaker identification, and sound source localization, among many others. For denoising the speech signals degraded with the many kinds and possibly different levels of noise, several algorithms have been proposed during the past decades, with recent proposals based on deep learning presented as state-of-the-art, in particular those based on Long Short-Term Memory Networks (LSTM and Bidirectional-LSMT). In this work, a comparative study on different transfer learning strategies for reducing training time and increase the effectiveness of this kind of network is presented. The reduction in training time is one of the most critical challenges due to the high computational cost of training LSTM and BLSTM. Those strategies arose from the different options to initialize the networks, using clean or noisy information of several types. Results show the convenience of transferring information from a single case of denoising network to the rest, with a significant reduction in training time and denoising capabilities of the BLSTM networks.

show abstract

Enhancing Speech Recorded from a Wearable Sensor Using a Collection of Autoencoders

Cited by 3 publications

References 37 publications

Integración de actividades de investigación y acción social en la docencia de Ingeniería Eléctrica

Integración de actividades de investigación y acción social en la docencia de Ingeniería Eléctrica

Exploring the potential of an audio application for teaching AI-based classification methods to a wider audience

Assessing the effectiveness of transfer learning strategies in BLSTM networks for speech fenoising

Contact Info

Product

Resources

About