A neural network model of hidden Markov model applied to the auditory periphery for speech processing and recognition

Presents a neural network model called Hidden Markov Model (HMM). It is applied to analysis and recognition of the response of the auditory periphery to speech stimulation. Based on the reports from some scientists, the electric discharge rate of the auditory nerve and the otoacoustic emissions of the auditory periphery underlying speech stimulation are related. This relation supports the idea that the response of the auditory periphery to speech is a random procedure of double layers. For the short-time response, the procedure is stationary time-invariant. For a long sequence response, the response procedure consists of many short-time states and the transition from one state to the next state is governed by a set of transition probabilities. The procedure of recognition is to know which observation vector is more matched with the codebook. The codebook is governed by a set of output probabilities. The Hidden Markov Model is just an appropriate algorithm to represent the random procedure of double layers. Moreover, in order to reduce the number of input parameters of HMM neural network, the vector quantization (VQ) is applied to converse the characteristic vectors to the observation values. For the authors´ experiment, a programmable device used to measure and process the response of the auditory periphery is developed in the authors´ laboratory. The device not only can detect transient otoacoustic emissions (TEOAE) and distortion product otoacoustic emissions (DPOAE), but also can synthesize any stimulation, such as speech, and then receive the generated response in the auditory periphery, and finally automatically recognize the response. The primary results presented here show that HMM has some potential possibilities in the application to speech processing and recognition by the auditory periphery. Therefore, further research will be of benefit to design hearing aids and a front-end for speech recognition, etc