Optimization Study Of HE-AAC Audio Decoder Filter Bank On ARM

High efficiency advanced audio decoding (HE-AAC) is the most advanced technology of the perceptual audio decoding, which can deliver better audio quality, higher compression rates, more channels and varieties of sampling rates, improve at least30%than the efficiency of advanced audio decoding (AAC). It has been widely used in quite a lot fields. And today, the consumer electronic products with HE-AAC decoder have become very hot.HE-AAC audio decoder have three filter bank modules, respectively is IMDCT, analysis filter bank and synthesis filter bank. The basic function of filter bank is allowed a portion frequency of the signal smoothly through, and the other part frequency of it is large inhibition. The filter bank accounts for more than60%of the computational complexity the whole decoding system, has very important role. With decoding technology improve continuously, the research of filter bank is also more and more deeply.This paper mainly studies the optimization of HE-AAC audio decoder filter bank based on ARM. Firstly, this paper research and analysis original algorithm algorithm of the filter bank. Then respectively adopt the base4IFFT algorithm, DCT-Ⅲ algorithm, DCT-Ⅲ and DCT-Ⅱ combination algorithm to optimize the IMDCT module, analysis filter bank and synthesis filter bank. Secondly, according to the characters of ARM platform, we write efficient fixed point arithmetic code based on the principle of fixed. We using the advantage of NEON grammar, realized respectively the above three optimization algorithm and improved filter bank in distribution of the register, function parameters, circulation variable, function calls, divide, multiply shift operation, general data processing, load/store data. Filter bank after NEON grammar optimized code size change smaller and through Conformance test of the HE-AAC audio standard, reduce memory requirement and improve decoding efficiency.