Evaluation of Out of Breath Speech Using Machine Learning Approaches

Abstract

Stress alters the speech production mechanism. Factors like emotion, cognitive load, pathology, noisy condition (Lombard effect), physical load, sleep deprivation, etc., affect speech production. Among these, speech under emotional, noisy, and pathological conditions are investigated extensively. Little light has been shed on speech under physical load conditions, called out-of-breath speech. Such evaluation of out-of-breath conditions can be used in context-aware speech interfaces to estimate the workload level, exercise intensity of an athlete, and physical fitness of a person.