Single-channel speech enhancement using implicit Wiener filter for high-quality speech communication

Abstract Speech enables easy human-to-human communication as well human-to-machine interaction. However, the quality of speech degrades due to background noise in environment, such drone embedded during search and rescue operations. Similarly, helicopter noise, airplane station reduce speech. enhancement algorithms resulting a crystal clear noise-free conversation. For many applications, it is also necessary process these noisy signals at edge node level. Thus, we propose implicit Wiener filter-based algorithm for using computing system. In proposed algorithm, first order recursive equation used estimate noise. The performance evaluated two utterances, one uttered by male speaker other female speaker. Both utterances are degraded different types non-stationary noises exhibition, station, drone, helicopter, airplane, white Gaussian stationary with signal-to-noise ratios. Further, compare conventional spectral subtraction algorithm. Performance evaluations objective measures demonstrate that outperforms estimating clean from Finally, implement addition on Raspberry Pi 4 Model B, which low power device.