Tesis doctorales: TESEO

Ayuda Salir

Ficha de Tesis

Imprimir

Ministerio de Universidades - Gobierno de España

Título: NEUROMORPHIC AUDIO PROCESSING THROUGH REAL-TIME EMBEDDED SPIKING NEURAL NETWORKS
Autor: Domínguez Morales, Juan Pedro
Universidad: Universidad de Sevilla
Departamento: Arquitectura y tecnología de computadores
Fecha de Lectura: 03/12/2018
Mención Europeo / Mención Internacional: Concedido
Programa de doctorado: Programa de Doctorado en Ingeniería Informática por la Universidad de Sevilla
Dirección:
- Jimenez Fernandez, Angel Francisco (DIRECTOR)
- Domínguez Morales, Manuel Jesús (CODIRECTOR)
- VICENTE DIAZ, SATURNINO (TUTOR/PONENTE)
Tribunal:
- Civit Balcells, Anton (PRESIDENTE)
- Linares Barranco, Alejandro (SECRETARIO)
- MORGADO ESTEVEZ, ARTURO (VOCAL)
- CABELLO PARDOS, ENRIQUE (VOCAL)
- Plana Cabrera, Luis (VOCAL)
Descriptores:
- REDES NEURONALES
- RECONOCIMIENTO Y SINTETIZACION DE HABLA
- SOFTWARE
- INFORMATICA
Localización: ESCUELA INTERNACIONAL DE DOCTORADO (EIDUS)

434415_1141610.pdf

Marcador: https://www.educacion.gob.es/teseo/mostrarRef.do?ref=1712415Añadir a favoritos.
Resumen:
Neuromorphic engineering is a discipline that studies, designs and implements hardware and software with the aim of mimicking the way in which nervous systems work, focusing its main inspiration on how the brain solves complex problems easily. Audio processing and speech recognition have become important tasks to improve the human-machine interface. Taking into account the limitations of current automatic speech recognition systems, like non-real time cloud-based solutions or power demand, recent interest for neural networks and bio-inspired systems has motivated the implementation of new techniques. Among them, a combination of spiking neural networks and neuromorphic auditory sensors offer an alternative to carry out the human-like speech processing task. In this work, novel speech recognition and audio processing systems based on a spiking artificial cochlea and neural networks are proposed and implemented, including a software tool for post-processing the information obtained from spiking cochleae, a hardware implementation of an audio samples classifier using spiking neural networks, a Deep Learning-based system for the recognition of heart murmurs that improves current state-of-the-art solutions, and a real-time speech recognition approach for voice commands classification using Deep Learning mechanisms and spiking convolutional neural networks implemented in hardware. The performance and efficiency of these systems have been evaluated, obtaining conclusions and proposing improvements for future works.

Imprimir

Modificar Consulta Nueva Consulta

Versión 4.4