Subject-in-the-loop system for auditory feedback based voice retraining

The paper presents the development of an experimental setup in a subject-in-the-loop configuration for voice retraining. The goal is to provide means for modifying human voice by feeding back to the subject the computer altered signal as a mental reference for voice retraining. This system can be used for recuperating voice quality after accidents or illnesses that affected normal voice. A main challenge in the design of such a system is the achievement of a latency of the reception of the modified voice signal that is so small that is not perceived by the subject versus the voice output production. Various alternatives of the development of such a computed based voice manipulation system were investigated. The experimental setup validity was verified in real-time tests.