Español

Abstract

This study examines the recovery quality of natural and synthetic Spanish voice signals using compressive sampling techniques. We collected samples from male voices, including an elderly adult, an adult, a child, and a synthetic voice. Various sparse representation techniques were employed, and quality metrics such as signal-to-noise ratio, mean squared error, and correlation coefficient were evaluated to compare the original signals with the recovered ones at different compression factors. Naturally, the results indicate that the recovery quality of natural voice signals surpasses that of synthetic signals, and the number of samples also has a significant impact on this aspect. However, the developed tool clearly demonstrates the properties and limitations of the recovered synthetic voice.