Design And Implementation Of AAC Audio Encoding Based On Load Balancing

With the growing demand for embedded multimedia applications, MEPG-4 AAC is widely used as a kind of audio coding standard for having high compression performance and providing better quality audio coding with low bit rate. However, without optimization, AAC real-time software encoding is difficult to achieve in embedded device for complexity algorithm. Therefore, the purpose of this thesis is to achieve AAC real-time encoding by algorithm optimization and load balancing, according to the characteristics of embedded device and mobile phone in the application of network video monitoring.For AAC encoding is highly modular, psychoacoustics model was selected as the most time consuming part. The complexity of the psychoacoustics model was reduced by separating window type judgment from psychoacoustics model and using spectrum value calculated by MDCT. According to the characteristics of SEP6300A, the assembly optimization was performed for the code compiled by C language compiler still has high redundancy and room to improve. According to the feature of embedded device and mobile phone under the application scenario, part of the audio data were encoded on mobile phone to reduce the encoding time in embedded device based on load balancing.The AAC real-time encoding was achieved by algorithm optimization and load balancing. The result showed that all the requirements were reached:the encoding time in embedded device was reduced by about 80%; the AAC real-time encoding was achieved in embedded device and mobile phone; the CPU usage of mobile phone was less than 10%; the audio quality calculated by PESQ was more than 3.