A kind of high-performance audio signal processor, FlexEngine, has been developed by researchers from the ASIC and System department of The Institute of Microelectronics of Chinese Academy of Sciences （IMECAS）. FlexEngine bases on 16-bit ASIP (Application Specific Instruction Set Processor) architecture, adopting 5-7 pipelines and the Dual-Harvard memory architecture. For different audio application conditions, FlexEngine can customize audio repairation personally, according to some acceleration instructions such as butterfly, convolution, extraction and division. It mainly achieves mainly audio processing functions such as audio frequency division, audio compensation, noise elimination and echo cancellation. FlexEngine can meet the requirement of speech signal real-time processing at the frequency of 5MHz, with the power consumption of 600µW.
Basing on the achievement above, the team is developing a kind of novel heterogeneous multi-core audio signal processor platform: FlexEngine plays as master core, integrating reconfigurable ASIC audio function nodes by efficient interconnection network on the chips, which can provide flexible and universal audio application development environment. The team has mastered the ultra low-power technology of SoC(System on Chip), which can be widely applied in other fields of portable medical electronic application.

Fig.1. FlexEngine (Image by IMECAS)

Fig.2. Test Environment of FlexEngine (Image by IMECAS)