Preliminary analysis of cough sounds

Cough is an important symptom in many diseases and at times is the only major symptom to diagnose some particular ailments. Cough is the powerful mechanism of human body to clear the central airways. Analyzing the cough type, its intensity and sound, the medical experts can estimate enough details about the ailment and appropriate cure. Hence, it should be possible to estimate the cough type and the disease by analyzing the acoustic signal of cough sounds. This paper describes a database collected for the analysis of cough sounds. Preliminary analysis of cough sounds is carried out using spectrograms, since these provides the ground truth information about the changes in the instantaneous fundamental frequency and its harmonics. Various standard signal processing tools may be used further to extract the features and characterize the cough sounds. The cough sounds database also consist of simulated (i.e., voluntary) cough sounds and normal speech signals of the healthy speakers, so as to compare the changes in the acoustic characteristics of cough sounds from healthy to unhealthy state of speakers, and between wet and dry cough. Initial results appear to be encouraging.