Abstract
Four band noise vocoded speech sounds were used to understand speech processing system in the central nervous system. Temporal change in frequency (or spectral) information, such as formants and the fundamental frequency, has been believed to be essential components for speech information processing. However, recent studies revealed that four band noise vocoded speech sounds where speech signals were replaced by four bands of noise with preserved ultra-slow amplitude envelope (low pass filtered at 16 Hz) got very intelligible (greater than 80%) after a short training period. Even intonation could be easily discriminated, which is usually discriminated by a change in the fundamental frequency. These facts indicate we start to create plasticity by using central pathways, which were not conventionally used, and match the outputs of the pathways with the existing templates. That is why in the beginning these noise vocoded speech sounds were not intelligible at all. Here psychophysical studies were carried out with an aid of fMRI measurements. The outcome of the present research would contribute to understanding the central system for speech perception and recognition, understanding reference structure in the brain and constructing systems for training perceptually impaired people. This research was supported by Special Coordination Funds, and a grant to RCAST at Doshisha University from the Ministry of Education, Culture, Sports, Science and Technology of Japan. [Jpn J Physiol 54 Suppl:S24 (2004)]
