Enhancement of electronic artificial larynx speech by denoising

Electronic artificial larynx (EAL) is important speech rehabilitation prosthesis for laryngectomees. However, existing EAL have some drawbacks such as harsh, raucous and unpleasant sound. One of the major problems is radiation-noise. It is a major source of the degradation in acceptability and intelligibility of EAL speech. In this study, we restrained radiated noise using independent component analysis based adaptive noise canceling. An acoustic analysis of the denoised EAL speech revealed a significant reduction in the amount of background noise yet preserved the acoustic characteristics of the vocal output. Perceptual evaluation also indicated an obvious preference for denoised speech.