Testing three different Speech Synthesizers to acknowledge the advantages of DNN systems against HMM Methods

Autores/as

  • Carlos Angel Franco-Galvan Universidad Nacional Autónoma deMexico
  • José Abel Herrera-Camacho Laboratorio de Tecnologías del Lenguaje, UNAM

DOI:

https://doi.org/10.13053/cys-29-3-4517

Palabras clave:

Speech Synthesis, Voice Parameterization, Line Spectral Pair, Hidden Markov Models, Deep Neural Networks

Resumen

Abstract. This document reports MOS results after testing naturalness and expressiveness in three different speech synthesis systems. A first system is based on HMM, the second one combines HMM and DNN and the third one is solely based on DNN. According to the results, DNN systems outperform HMM systems in synthetic speech quality.

Biografía del autor/a

Carlos Angel Franco-Galvan, Universidad Nacional Autónoma deMexico

PhD Student Posgrado en Ciencia e Ingeniería de la Computación

José Abel Herrera-Camacho, Laboratorio de Tecnologías del Lenguaje, UNAM

Abel Herrera Camacho is head of Laboratorio de Tecnologías del Lenguaje FI-UNAM. He specializes in speech synthesis and recognition and has published work on the field for over 20 years.

Descargas

Publicado

2025-09-25