An Experimental Study on Speech Enhancement Based on a Combination of Wavelets and Deep Learning

Abstract

The purpose of speech enhancement is to improve the quality of speech signals degraded by noise, reverberation, or other artifacts that can affect the intelligibility, automatic recognition, or other attributes involved in speech technologies and telecommunications, among others. In such applications, it is essential to provide methods to enhance the signals to allow the understanding of the messages or adequate processing of the speech. For this purpose, during the past few decades, several techniques have been proposed and implemented for the abundance of possible conditions and applications. Recently, those methods based on deep learning seem to outperform previous proposals even on real-time processing. Among the new explorations found in the literature, the hybrid approaches have been presented as a possibility to extend the capacity of individual methods, and therefore increase their capacity for the applications. In this paper, we evaluate a hybrid approach that combines both deep learning and wavelet transformation. The extensive experimentation performed to select the proper wavelets and the training of neural networks allowed us to assess whether the hybrid approach is of benefit or not for the speech enhancement task under several types and levels of noise, providing relevant information for future implementations.

Detection of E. coli concentration levels using CSI-D+ handheld with UV-C fluorescence imaging and deep learning on leaf surfaces

“…Subsequently, these coefficients were utilized for image reconstruction and the extraction of denoised images post 2D inverse WT (IDWT) operations. The quality of denoised images were assessed by comparing them with the original noisy ones using a metric called peak signal-to-noise ratio (PSNR) [21], [23], [24].…”

Extending Radio Broadcasting Semantics through Adaptive Audio Segmentation Automations

“…Sound recognition plays an important role in most of the encountered audio and audiovisual pattern analysis cases, where related content is massively produced and uploaded (i.e., digital audio broadcasting, podcasts and web radio, but also video on demand (VoD), web-TV and multimodal UGC sharing in general). Specifically, there are various pattern recognition and semantic analysis tasks in the audio domain, including speech-music segmentation [8], genre recognition [10], speaker verification and voice diarization [11], speech enhancement [12,13], sound event detection [14], phoneme and speech recognition [15][16][17], as well as topic/story classification [18][19][20], sentiment analysis and opinion extraction [4,5,21], multiclass audio discrimination [22], environmental sound classification [23] and biomedical audio processing [24]. Audio broadcast is generally considered to be one of the most demanding recognition cases, where a large diversity of content types with many detection difficulties are implicated [1].…”