Acoustic voice markers
Speech features of sleepiness

The vocal markers most commonly used to study the influence of pathologies on the voice are acoustic descriptors of voice quality.

The most widely used software for extracting them is openSMILE which extracts over 6,000 different characteristics from a voice recording.

The main drawback of using this toolbox is the lack of explicability of the very many parameters proposed.

On the contrary, our approach is to design a small number of markers that can be explained and validated by sleep physicians who are not voice or signal processing specialists....