FPGA implementation of spectral analysis stage used in binary mask algorithm for hearing applications

This paper presents implementation of binary mask algorithm with focus on spectral analysis stage, which can be suitable for different hearing applications. FPGA implementation of spectral-analysis stage is provided, while classification and synthesis stages are implemented in Matlab. Three different types of noises at three different levels in dB are considered to evaluate performance of this algorithm. Results have shown considerable improvement in intelligibility which is measured with the help of short time objective intelligibility test. Average intelligibility obtained is 83.16%.