Yi-Hsuan Lee

A Vocoder for a Novel Cochlear Implant Stimulating Strategy Based on Virtual Channel Technology

Cochlear implant provides opportunity for profoundly hearing impairment patient to have a chance to hear sound again. However, the limited number of electrodes is in sufficient to provide enough hearing resolution for hearing impaired people, especially for application in tonal language and music. Virtual channel technology opens up the possibility to increase the hearing resolution under the limited electrodes available and improve and quality for tonal language and music. In this paper, vocoder implementation of a new speech strategy based on virtual channel technology is used to study the improvement. The test result shows improved understanding and quality with virtual channels over the traditional strategies, ex. CIS, especially for mandarin and music.

A Review of Stimulating Strategies for Cochlear Implants

Many animals use sound to communicate with each other, and hearing is particularly important for survival and reproduction. In species that use sound as a primary means of communication, their hearing is typically most acute for the range of pitches produced in calls and speech. Human is one such species and Fig. 1 shows a human ear consisting of the outer, middle and inner ear. The eardrum of an ear converts incoming acoustic pressure waves through the middle ear to the inner ear. In the inner ear the distribution of vibrations along the length of the basilar membrane is detected by hair cells. The location and intensity of these vibrations are transmitted to the brain by the auditory nerves. If the hair cells are damaged (as shown in Fig. 2(b)), the auditory system is unable to convert acoustic pressure waves to neural impulses, which results in hearing impairment. Damaged hair cells can subsequently lead to the degeneration of adjacent auditory neurons. If a large number of hair cells or auditory neurons are damaged or missing, the condition is called profound hearing impairment (Yost, 2000).

Channel interaction in cochlear implant acoustic models

Cochlear implants (CI) provide opportunities for people with profound hearing impairment to recover partial hearing. An acoustic CI model based on specific stimulating strategy can be used to study sounds processed in CI, and allows test subjects with normal hearing to evaluate stimulating strategy performance. However, there is still a significant performance difference between hearing test results based on acoustic CI models and from CI users. In this study we propose to use a SPREAD matrix, which incorporates channel interaction, created by the activating function profile based on finite element models of CI to improve the acoustic CI model. According to test results, acoustic CI models with SPREAD matrix based on finite element models significantly improves its match with clinical CI data and can be used to evaluate the performance of CI stimulating strategies and predict the hearing performance of CI users more accurately.

Speech Recognition for Cochlear Implant Users in the Noisy Environment: The Role of Envelope and Fine Structure

Sound signal can be decomposed to a slowly varying envelope cue and a rapidly varying fine structure cue. These cues can help people to sound perception, sound lateralization, speech recognition in noise, and so on. Cochlear implant can help people with hearing loss to hear the sound. However, there are still many restrictions with cochlear implant user and a gap between normal hearing and cochlear implant user. This study investigated the contribution of envelope and fine structure cues with different SNR and simulated cochlear implant in Taiwanese mandarin speech recognition in noise. Ten normal hearing subjects participated in this experiment. Our result shows sentence recognition in noise almost depends on the envelope cues, but the fine structure cues still have limited contributions.

Low Frequency Acoustic Information for Cochlear Implant Users with Hearing Aids in Contralateral Ear

Recent studies have reported speech intelligibility in noisy environment from cochlear implant (CI) users was reduced because the spectral resolution is not high enough. For the user who had only implanted one CI in unilateral ear, combined a hearing aid (HA) in contralateral ear, known as bimodal hearing, can improved the speech perception. This is partly true due to the residual low frequency acoustic information provided from HA can help CI users to segregate the target speech from noise. In this study, a vocoder was conducted to evaluate the effect of low frequency information, especially the fundamental frequency (F0), in recognizing Taiwanese Mandarin for bimodal CI users. The result showed that F0 was the main factor in low frequency acoustic information, and bimodal apparently enhance the performance of speech recognition in noise in Taiwanese Mandarin.

Acoustic Cochlear Implant Models Incorporating Electrical Field Interaction between Neighboring Electrodes

Cochlear implant (CI), which has been a com-mercial device for nearly thirty years, can provide opportuni-ties for patients with severe to profound hearing impairment to regain hearing. Sound processed in the CI can be modeled by acoustic CI models on specific stimulating strategies. Speech processed by acoustic CI models was used to test normal hear-ing test subjects and can be a basis to evaluate the perfor-mance of various stimulating strategies. Unfortunately, there is still a significant performance difference between hearing test result based on acoustic CI model with normal hearing test subjects with hearing tests from CI patients. There are a num-ber of explanations for this, but one important factor is that traditional acoustic models do not account for the electric field interaction between electrodes. For CI users, the electrical field interaction or current spread between electrodes distorts the speech and introduce noises. In order to model the electri-cal field interaction in the acoustic model, we proposed a new SPREAD model which is created by the activating function profile generated from a finite element model we developed. In previous SPREAD model, the current spread was fixed at a certain decay rate. In the proposed model, activating function and finite element model of a CI is introduced to improve the SPREAD matrix. The performance of the proposed SPREAD matrix with the acoustic CI model significantly improves its match with clinical CI data. This shows the acoustic CI model with the proposed SPREAD matrix can be a more accurate model to evaluate the performance of CI stimulating strategies.

Extracting Speech Signals using Independent Component Analysis

Independent component analysis (ICA) is the dominant method to resolve blind source separation (BSS) problem. In this article we conducted experiments to evaluate the separation performance of ICA for acoustic signals. Experiments results show that if we can find appropriate placement of microphones, applying ICA to hearing prostheses as pre-processing can help the wearer hear more clear sounds.