Convolutional Neural Networks to Enhance Coded Speech

2018/06/25 by Ziyue Zhao, Zhao, Ziyue, Huijun Liu +3 · 2 citations
Computer Science · Engineering · Neuroscience · #Acoustic Wave Phenomena Research #Artificial intelligence #Audio and Speech Processing (eess.AS) #Audio signal #Cepstrum #Codec #Computer science #Convolutional neural network #FOS: Computer and information sciences #FOS: Electrical engineering #Hearing Loss and Rehabilitation #Linear predictive coding #Noise reduction #PESQ #PSQM #Sound (cs.SD) #Speech and Audio Processing #Speech coding #Speech enhancement #Speech recognition #Telecommunications #Wideband audio #cs.SD #eess.AS #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.1806.09411

published in arXiv (Cornell University) (Cornell University) · More analysis are added for version 4

openalex publication_date 2018/06/25 · arxiv created 2019/01/24 · arxiv updated 2019/01/25 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/06

Abstract

Enhancing coded speech suffering from far-end acoustic background noise, quantization noise, and potentially transmission errors, is a challenging task. In this work we propose two postprocessing approaches applying convolutional neural networks (CNNs) either in the time domain or the cepstral domain to enhance the coded speech without any modification of the codecs. The time domain approach follows an end-to-end fashion, while the cepstral domain approach uses analysis-synthesis with cepstral domain features. The proposed postprocessors in both domains are evaluated for various narrowband and wideband speech codecs in a wide range of conditions. The proposed postprocessor improves speech quality (PESQ) by up to 0.25 MOS-LQO points for G.711, 0.30 points for G.726, 0.82 points for G.722, and 0.26 points for adaptive multirate wideband codec (AMR-WB). In a subjective CCR listening test, the proposed postprocessor on G.711-coded speech exceeds the speech quality of an ITU-T-standardized postfilter by 0.36 CMOS points, and obtains a clear preference of 1.77 CMOS points compared to legacy G.711, even better than uncoded speech with statistical significance. The source code for the cepstral domain approach to enhance G.711-coded speech is made available.

Citations

• Image Super-Resolution Using Deep Convolutional Networks
• An algorithm for the machine calculation of complex Fourier series
• Capacity of a Burst-Noise Channel
• On the Computation of the Discrete Cosine Transform
• Speech enhancement using a minimum mean-square error log-spectral amplitude estimator
• Raw Waveform-based Speech Enhancement by Fully Convolutional Networks
• End-to-End Waveform Utterance Enhancement for Direct Evaluation Metrics Optimization by Fully Convolutional Neural Networks
• Learning Deconvolution Network for Semantic Segmentation

Cited by

• Voice-ENHANCE: Speech Restoration using a Diffusion-based Voice Conversion Framework

Related

• LACE: A light-weight, causal model for enhancing coded speech through adaptive convolutions
• PostGAN: A GAN-Based Post-Processor to Enhance the Quality of Coded Speech
• AMRConvNet: AMR-Coded Speech Enhancement Using Convolutional Neural\n Networks
• End-to-End Optimized Speech Coding with Deep Neural Networks
• A DNN Based Post-Filter to Enhance the Quality of Coded Speech in MDCT Domain