Design of cochlear implant device for transmitting voice pitch information in speech sound of asian languages

Resolution of the fundamental frequency of speech sound required for the design of a speech processor of a cochlear implant device is investigated, with special regard to transmitting voice pitch information in Asian languages. Clinical application of the cochlear implant has spread rapidly in recent years to Asian countries where a variety of languages having different voice pitch information from English and other European languages are spoken. The perceptually acceptable area and required resolution of duration and fundamental frequency is estimated on a twodimensional chart consisting of logarithmic time and frequency scales, based on the typical voice pitch contours of Japanese word accent and Chinese syllabic tone. As a result, it is shown that much finer quantizing and time sampling for the change in fundamental frequency is required compared with sentence intonation and emphasis common to other languages. It is also shown that the amount of information conveyed by combined use of lipreading with a cochlear implant is not sufficient for supplementing the voice pitch information. A possible way of transmitting such voice pitch information by transmission of the waveform of speech sound directly to the auditory area of cortex, where the waveform is reconstructed and voice pitch is extracted, is discussed. 1. COCHLEAR IMPLANT 1.1. Processes involved in the device In a cochlear implant device (or an artificial inner ear device), speech sound is converted into an electrical signal and is transmitted to the electrode implanted in the inner ear, in order to directly stimulate the auditory nerve of a deaf person. The processes involved in a cochlear implant device are schematized in Figure 1 (References 1 and 2). Speech sound through a microphone and a pre-processor is input to a bank of band-pass filters, which are assigned from low to high frequency components of speech sound waveform. The output, after being rectified and smoothed by low pass filters, is fed to the speech signal processor, where the peak channels are enhanced, and formants and voice pitch are extracted. The information from the filters is encoded to map to the electrodes, transmitted transcutaneously from the radio frequency sender to receiver, and decoded and fed to the electrode array. RADIO FREQUENCY SENDER TRANSCUTANEOUS TRANSMISSION RADIO FREQUENCY RECEIVER PROCESSES INVOLVED IN THE COCHLEAR IMPLANT DEVICE MICROPHONE / PRE-PROCESSOR SPEECH SIGNAL PROCESSOR ENHANCE PEAK CHANNELS EXTRACT FORMANTS / VOICE PITCH BANDPASS FILTERS / RECTIFIERS / LOWPASS FILTERS ENCODER MAPPING FROM FILTERS TO ELECTRODES