3D Audio Engine Porting And Optimization On ARM11

3D Positional Audio Technology is one of the hottest research achievements in embedded multimedia domain. By implementing HRTF (Head Related Transfer Function) and Reverberation,3D Audio Engine has the capability of creating 3D positional audio through stereo headset or dual-speaker, and provides end user an amazing audio experience which they never heard on other handset. Due to the increasing requirement from embedded market, the audio professionals like Dolby, SRS, Sonaptic all released 3D positional audio engine based on the 3D audio technology. Meanwhile the mainstream embedded product vendor especially mobile phone manufactories also dedicate themselves into research and development of 3D audio enabled phones by cooperating with professional 3D audio company, which is coming to be a most popular technical highlight in mobile market. Based on the C fixed-point code from Motorola, this paper is going to show how we port, optimize and integrate 3D Audio Engine on ARM 11 platform.To optimize the complex operations and dense computations applied in 3D Audio Engine, this thesis uses some new techniques of ARMv6 version, such as SIMD etc. After analyzing MIPS hotspots of core algorithm, I took the following optimizing strategies:loop unroll, register reorganization, parallel computing, and pipeline optimization.So far we have successfully completed porting and optimizing the 3D Audio Engine on ARM11 with conformance test passed. The average MCPS of 1 sound source is reduced from 243.31 to 13.81 and the total memory is within 60 Kbytes, which completely obtains the project's goals. By 2011 Motorola has released 10+3D audio capable mobile phones in North America and created unique multimedia experience in music playback,3D gaming,3D ringtone etc.