A study of sensory substitute systems based on biomedical acoustics

本研究的目的是建立基于生物医学声学的感觉替代系统。研究结果如下：1.提出了一种新的聋人触觉辅助方法。在这种方法中，语音频谱模式在指尖上被扫描为元音的凸模式和辅音的振动模式。实验证明，该方法可以显著提高语音识别率.针对喉功能丧失的患者，设计了一种具有呼气压力控制语调功能的电子喉。这种假肢被发现是非常有益的，在生产正常的发声讲话。这是与一家电子公司合作开发的实用软件。设计了一种刺激耳蜗神经的耳鸣抑制器，并研究了电刺激对耳蜗的影响。我们为7例患者植入了耳鸣抑制器。实验证明，无论耳鸣何时出现，它们都能自行抑制耳鸣.通过心理物理学实验，阐明了盲人障碍感产生的机制。障碍感被认为是由以下现象解释：前效应，从两个耳朵获得的两个声音信号的相关性，和着色。研究结果将应用于盲人康复.为后天失聪者设计了一台语音打字机。在语音打字机中，68个日语单音节语音被转换成日语字母。其响应时间为0.2秒，识别率约为96%。语音打字机的原型大约在17年前开发出来。然而，原型太贵了，聋人买不起。通过本研究，我们可以改进小型廉价的语音打字机的设计，使其实用化。

The purpose of this research is to create sensory substitute systems based on biomedical acoustics. The research results are shows as follows.1. A new tactile aide method for the deaf was proposed. In this method, speech spectral patterns were swept on a fingertip as convex patterns for vowels and vibration patterns for consonants. This method was proven to dramatically increase the speech recognition rate.2. An electro-larynx which has a function of intonation control by using expiration pressure was designed for people who have lost their larynx. This prosthesis was found to be very beneficial in producing normal sounding speech. This was developed for practical use in cooperation with an electronic company.3. A tinnitus suppresser which stimulates the cochlear nerves was designed and some effects of the electrical stimulation to the cochlea were made clear. We have implanted the tinnitus suppresser for seven patients. It was proven that they could suppress the tinnitus by themselves whenever the tinnitus appeared.4. The mechanism of obstacle sense of the blind was clarified based on psychophysical experiments. The obstacle sense was found to be explained by the following phenomena : preceding effect, correlation of two sound signals obtained from both ears, and coloration. The findings will be applied to rehabilitation for the blind.5. A voice typewriter was designed for the acquired deaf. In the voice typewriter, 68 Japanese monosyllabic voices were converted into Japanese letters. It's response time was 0.2 sec and the recognition rate was about 96%. The prototype of the voice typewriter was developed about 17 years ago. However, the prototype was too expensive for the deaf to buy. In the research, we could improve on the design of a small and cheap voice typewriter for practical use.