With the higher need of communication efficiency and quality, audio vocoder technique also immediately develops. From the waveform coding and parameter coding in earlier period to mixed coding and perceptual coding till now. So how to accurately evaluate the performance of these vocoders is of the following work. The assessment of audio quality can be classified into two broad categories: Subjective assessment and Objective assessment. The sound is felt by people finally, so subjective assessment is always the effective method. But its operating procedure and experimental condition are often more complex, resulting in its application has some constraints. In contrast, the objective evaluation methods have the character of real-time processing and good reproducibility.Perceptual evaluation of audio quality is proposed by the ITU. It can evaluate the performance of the vocoder and other audio system. It is based on psychoacoustic model, through the time-frequency transformation, the frequency grouping, the masking calculation and the harmonic analysis to simulate the process that how the human ear perceives sound. It is the best algorithm to match with the subjective assessment (SDG) so far.This thesis presents the principle and characteristic of the subjective assessment and objective assessment firstly. Then the writer analyses the psychoacoustic model in two aspects of the human auditory system and the perceptive characteristic of the human ear, which are also the main content of the PEAQ algorithm. The thesis also has an in-depth discussion and some improvement for the important module in the later chapters. The writer analyses the achieved results finally. |