Electrical stimulation of the auditory system provides considerable speech information to deaf patients, even to those using only a single electrode. We propose a series of psychophysical measures to quantitatively define basic perceptual capabilities for detecting and discriminating simple and complex temporal patterns with electrical and acoustic stimulation. A unique feature of this proposal is access to patients with different cochlear implants (CI) and patients with implants on the cochlear nucleus (ABI: auditory brainstem implants). Implant patients have no mechanism for spectral analysis of electrical signals and so can only receive time-intensity or envelope information. However, it is not clear what type of envelope information is important for speech recognition and how to present this information efficiently to implanted patients. Our hypothesis is that low-frequency temporal information and amplitude information reflecting the envelope of the auditory stimulus are perceived normally by implanted patients. A secondary hypothesis is that simple psychophysical tasks do not relate to speech discrimination because they measure basic capability in optimal conditions rather than measuring capability in more complex and dynamic conditions found in speech. The long-term objective is to understand the relation between speech recognition and psychophysical capability on time-intensity patterns.
The specific aims of this research are to (1) design psychophysical tasks to measure detection and discrimination performance for stimuli whose time-intensity patterns reflect important phonetic distinctions in speech, (2) measure the categorization of these temporal patterns in normal and electrically stimulated listeners, and (3) design and implement single- and multi- channel speech processors that preserve salient features of the time- intensity pattern. The experimental design and method will be to measure detection and discrimination of temporal patterns by objective 2AFC tracking methods. The temporal patterns will be selected to represent temporal features of speech signals that are known to be important in phonetic distinctions. We will measure categorization of temporal patterns using a traditional labeling task with stimuli distributed along continuum spanning the categories. Single and two-channel speech processors will be designed based on the psychophysical studies to preserve timely-intensity patterns that are important in distinguishing phonetic information. Consonant, vowel, and connected speech recognition will be measured for several processors and the results compared to predictions based on psychophysical data.
Showing the most recent 10 out of 43 publications