Speech Recognition and Synthesis Algorithm for Digital Hearing Aids under Background Noise

The difficulty of speech comprehension under background noise greatly influenced the use of hearing aids. To address this problem, this paper proposes a new hearing aids algorithm based on speech recognition and synthesis. This method is based on the pure speech to build a parameters database. Under the real noisy scene, implement speech recognition for the input speech, and then extract the corresponding parameters according to the results of the recognition and synthesize the pure speech. To improve the noise robustness of speech recognition system, we use a newly proposed kernel power flow orientation coefficients (KPOCs) as the characteristic parameters for training and testing. Because the speech synthesis is based on the parameters of pure speech, the output speech almost does not contain noise, and the difficulty of speech comprehension has great improvement.