ACOUSTIC PROCESSING OF SPEECH TO IMPROVE ELECTROLARYNX COMMUNICATION

DESCRIPTION (provided by applicant): Over half of laryngectomy patients use an electrolarynx (EL) to communicate, but current EL devices produce speech that has poor quality ("non-human sounding") and reduced intelligibility. The acoustic deficits in EL speech inhibit the ability of laryngectomy patients to communicate, thus reducing their functional capability and quality of life. The long-term goal of this project is to develop real-time speech processing technology to remedy the acoustic deficits in EL speech and thereby improve EL communication and quality of life for laryngectomy patients. Phase I will build on our preliminary studies that showed a significant improvement in the naturalness of EL speech with the addition of F0 variation, and demonstrated the feasibility of co-varying F0 based on modulations in the amplitude of the EL speech signal. An analysis-by-synthesis approach and perceptual experiments will be used to determine whether amplitude-based control of F0 can significantly improve the sound quality of EL speech, as well as restore the reduced capability that EL users have to signal linguistic contrasts involving intonation and contrastive stress. Given positive results in Phase I, Phase II will include development of a prototype (portable) DSP-based hardware platform for implementing the F0 control algorithm in real time. Further experiments utilizing analysis- by-synthesis and perceptual evaluation will be performed to identify additional acoustic deficits in EL speech (deficits in voice source, vocal tract resonance parameters, etc.) that can be corrected algorithmically to further improve sound quality and intelligibility. The ultimate goal of Phase II is to produce a prototype portable system would allow an EL user to communicate in real time using the newly developed speech acoustic enhancements either on a telephone or in a face-to-face situation. Of the approximately 1000 patients per year who undergo laryngectomy in the United States, more than half of them use an electrolarynx for communication post-laryngectomy. Unfortunately, EL speech contains persistent acoustic deficits that negatively affect intelligibility and social interaction. The proposed research will use an analysis-by-synthesis approach and perceptual experiments to determine whether amplitude-based control of fundamental frequency can significantly improve the sound quality of EL speech.