English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Despite significant efforts over the last few years to build a robustautomatic speech recognition (ASR) system for different acoustic settings, theperformance of the current state-of-the-art technologies significantly degradesin noisy reverberant environments.  Convolutional Neural Networks (CNNs) have been successfully used to achievesubstantial improvements in many speech processing applications includingdistant speech recognition (DSR). However, standard CNN architectures were notefficient in capturing long-term speech dynamics, which are essential in thedesign of a robust DSR system. In the present study, we address this issue byinvestigating variants of large receptive field CNNs (LRF-CNNs) which includedeeply recursive networks, dilated convolutional neural networks, and stackedhourglass networks. To compare the efficacy of the aforementioned architectureswith the standard CNN for Wall Street Journal (WSJ) corpus, we use a hybridDNN-HMM based speech recognition system. We extend the study to evaluate thesystem performances for distant speech simulated using realistic room impulseresponses (RIRs). Our experiments show that with fixed number of parametersacross all architectures, the large receptive field networks show consistentimprovements over the standard CNNs for distant speech. Amongst the exploredLRF-CNNs, stacked hourglass network has shown improvements with a 8.9% relativereduction in word error rate (WER) and 10.7% relative improvement in frameaccuracy compared to the standard CNNs for distant simulated speech signals.

Analyzing Large Receptive Field Convolutional Networks for Distant Speech Recognition