Yoshiyuki Yoshizumi

A study of signal‐to‐noise ratio conditions for measuring the intelligibility of hearing aids in noisy live environments

The authors studied a method to measure the intelligibility of directional hearing aids using a dummy head in noisy environments. The method is based on ISO recommendations (ISO 8353‐2, ISO 8353‐3). However, it is sometimes difficult to meet the quasi‐free‐field conditions. Therefore, a method for approximate measurement of intelligibility, by which the equivalent results are obtained in noisy live environments, is recommended. To simulate noisy environments, a multitalker (MULT) noise or a simulated multitalker noise was simultaneously and noncorrelatively put out through four loudspeakers in the live testing room. The MULT noise was composed of eight voices (four males and four females). The experiment showed that mixed voices of eight to ten persons were adequate for interference of the test speech words. The 50% SNR specification, meaning signal‐to‐noise ratio that gives intelligibility of 50% correct, was introduced into the measurement system. Intelligibility was measured under various SNR condition...

The design and evaluation of a noise suppression system for custom‐made digital hearing aids

A noise suppression system, which has the inbuilt feature to adapt to the local environment, for custom‐made digital hearing aids has been developed. The signal process is based on spectral subtraction technology with 32 bands in the frequency domain. The signal‐to‐noise ratio was improved by an average of 7.5 dB in rms level. Its effectiveness was evaluated using the seven‐point rating scale method for nine normal‐hearing subjects and six hearing impaired subjects. Test speech is presented under three different environments, which consist of a car, a duct, and a crowded place. As a result of the assessment, the same tendency is observed between normal and hearing‐impaired subjects and there were two good points that were found. First, the noise suppression system makes them more comfortable to listen to speech with the statistical significance of p=0.05. Second, the noisiness is reduced with the statistical significance of p=0.01. It should be noted that there is a scattering of the plotted data; however...