Deep Vocoder: Low Bit Rate Compression of Speech with Deep Autoencoder

TL;DR

This paper introduces Deep Vocoder, a low bit rate speech compression method using deep autoencoders and perceptually optimized vector quantization, achieving significant improvements in speech quality metrics.

Contribution

It proposes a novel end-to-end speech compression framework combining deep autoencoders with analysis-by-synthesis vector quantization optimized for perceptual quality.

Findings

Achieves higher PESQ scores at 2400 and 1200 bit/s.

Outperforms conventional SQ- and VQ-based codecs.

Demonstrates improved spectral and intelligibility metrics.

Abstract

Inspired by the success of deep neural networks (DNNs) in speech processing, this paper presents Deep Vocoder, a direct end-to-end low bit rate speech compression method with deep autoencoder (DAE). In Deep Vocoder, DAE is used for extracting the latent representing features (LRFs) of speech, which are then efficiently quantized by an analysis-by-synthesis vector quantization (AbS VQ) method. AbS VQ aims to minimize the perceptual spectral reconstruction distortion rather than the distortion of LRFs vector itself. Also, a suboptimal codebook searching technique is proposed to further reduce the computational complexity. Experimental results demonstrate that Deep Vocoder yields substantial improvements in terms of frequency-weighted segmental SNR, STOI and PESQ score when compared to the output of the conventional SQ- or VQ-based codec. The yielded PESQ score over the TIMIT corpus is 3.34 and 3.08 for speech coding at 2400 bit/s and 1200 bit/s, respectively.