A DNN Based Adaptive Filter for Speech Enhancement

The challenge of voice improvement in signal processing is enhancing the quality of speech signals recorded in noisy or unreliable environments. Digital communications, speech preparation for hearing aids, and recognition of speech all work better when speech augmentation is used. Enhancement of speech results in improving the transparency, ease of understanding and sound attributes of messages. Background noise is muted through speech enhancement. Numerous speech processing applications have emerged as a result of speech’s practicality. In this paper, a DNN-based adaptive filter is used to improve the speech quality. DNN estimates filter coefficients. The proposed DNN-based algorithm works well. It accurately estimates filter coefficients from clamorous speech. DNNs can learn well. This enables the proposed algorithm to improve speech quality and intelligibility.