Snore Recognition Using a Reduced Set of Spectral Features

ALBORNOZ, M. E.; BUGNON, L. A. ; MARTÍNEZ, C. A.

Mar del Plata

Congreso; XVII Reunión de trabajo en Procesamiento de la Información y Control; 2017

ICYTE

Snoring affects the sleep quality of the snorer itselfand its social circle. Some types of snoring are related to sleepapnea, which leads to sleepiness during the day and to severalhealth risks. Thus automatic detection of the different types ofsnoring may lead to more specific diagnosis and consequenttreatment. In this work, we propose to use a reduced set ofspeech related features that includes spectral information, Mel-Frequency Cepstral Coefficients (MFCCs), prosodic values andbioinspired information. Extreme Learning Machines (ELM)are proposed to learn on the non-linear feature set. A well-known classifier as Support Vector Machines (SVM) is used asbaseline. Several configurations for the feature sets and the ELMwere evaluated. The bioinspired information shows promisingresults on the Munich-Passau Snore Sound Corpus (MPSSC)with respect to the baseline performance on the developmentpartition.