A Research Of Auditory Attention Model In Speech Quality Assessment

As people’s expectations rising, the audio content tends to be more and more diversified and complicated. Psychological research shows that people in the complex sound field environment selective auditory concern mechanisms. Existing voice quality evaluation method based only on acoustic masking model and the underlying acoustic significant characteristics, there is a big difference with the real subjective listening experience, and therefore need to seek more accurate and reliable voice quality evaluation method. This thesis is psychological chose to focus mechanism to introduce voice quality evaluation system, in the subjective and objective evaluation methods to establish auditory attention model, makes the evaluation results are more consistent with human auditory characteristics.The subjective evaluation uses the existing MUSHRA platform. Concerned about the model hearing in the subjective evaluation, using the method of combining the typical environment and concerned about the voice, and property division of the auditory characteristics, evaluation of voice quality from different angles to solve the interference problem of the subjective individual differences and improve the accuracy of the subjective evaluation Finally, using the method of regression analysis to calculate the properties of the weighted parameter estimates.Objective evaluation using the existing PEAQ platform, hearing concern is the interaction of top-down and bottom-up two mechanisms, the typical environment and the underlying top-down acoustic characteristic parameters associated method, for the different environments in the background the sound differences to select the different parameters of the underlying acoustic characteristics, the use of regression analysis to fit the main objective evaluation of results, and ultimately establish a model of auditory attention in the objective evaluation of voice quality.This thesis is to establish auditory attention model, not only effectively reduce the subjective assessment of individual differences of voice quality disturbances, so that the evaluation is more true, accurate and objective evaluation method based on a typical environment, making the degree of correlation of objective method evaluation results more high.