A novel hardware architecture for Noise Cancellation based on adaptive speech probability density function

A novel hardware architecture for Noise Cancellation (NC) based on adaptive speech probability density function (PDF) is proposed in this paper. The used PDF has two adaptive shape parameters which affect the quality of enhanced speech. Noise can be efficiently suppressed by estimating properly these parameters so that the adaptive speech PDF shape fits to the real speech PDF one. A novel hardware architecture is proposed based on adaptive speech PDF algorithm which shows good results in implementation. Proposed hardware system with stage-pineline technique has achieved a maximum clock frequency of 127 MHz and processing latency is 8.6 μs, this enables noise cancellation processing can happen in real-time with very low clock frequency. Hardware scheduling technique is also applied to reduce operators for getting low-resource. A parameter controller configured by system firmware is also implemented to make NC system more flexible. Experimental results show that proposed NC system can suppress noise in real-time with high quality of enhanced speech.