SPEECH PROCESSORS FOR AUDITORY PROSTHESES

点击翻译按钮获取中文摘要

The objective of this contract is to design, develop and evaluate both laboratory based and wearable speech processors for use in auditory prostheses. An essential component of all auditory prostheses is the speech processor whose function is to convert the wide bandwidth, large dynamic range electrical signal from a microphone to a signal or set of signals for driving the individual electrical implant stimulators in a manner to optimize recognition of speech and environmental sounds by the implant user. Considerable progress has been made in understanding the information that is being conveyed to the remaining auditory nerve fibers and the cochlear nucleus during electrical stimulation with electrodes in the scala tympani and on the surface of the cochlear nucleus. Based in part on these findings, there have been significant improvements in speech processors and consequently in speech recognition by deaf implant users. It has been found that many, but not all, multichannel implant patients benefit from non-simultaneous application of stimuli to their different electrodes. Also, some subjects demonstrate improved understanding of speech when activation of their electrodes is accomplished by cycling through them at a high rate, i.e. faster than individual auditory nerve fibers can follow. Another important factor appears to be the loudness growth transfer function in the speech processor. The test score results from many of these research subjects have shown significant increases over scores achieved with their conventional commercial processors even though they have had only a few hours of practice with the experimental processors. In addition, portable digital speech processors have been designed and are being evaluated using some of the new speech processing strategies. There are several different types of single and multichannel auditory prostheses being used in human subjects. These include both unilateral and bilateral cochlear implants as well as cochlear nucleus implants. Each of these is based on different design philosophies and each has certain advantages/limitations. Since optimal speech processor design depends, among other things, upon both the implanted electrode design and the remaining auditory nervous system of the subject, it is important in designing speech processors to do so for specific implant designs and to test them with a representative cross section of users. Most speech processor evaluation has been done in low background noise environments. This work is being extended to permit the testing and evaluation of additional speech processing techniques in previously implanted patients under a range of signal-to-noise conditions with emphasis on subjects considered poor users with their commercial speech processors. Continued emphasis is being placed on converting some of the software-based speech processing systems into wearable versions for evaluation of the devices under conditions of "daily living" and the effects of long-term learning.