“Application of EαNets to feature recognitionof articulation manner in knowledge-based automatic speech recognition

Sabato M. Siniscalchi, Jinyu Li, and et. al

Abstract

Speech recognition has become common in many application domains. Incorporating acoustic-phonetic knowledge into Automatic Speech Recognition (ASR) systems design has been proven a viable approach to rise ASR accuracy. Manner of articulation attributes such as vowel, stop, fricative, approximant, nasal, and silence are examples of such knowledge. Neural networks have already been used successfully as detectors for manner of articulation attributes starting from representations of speech signal frames. In this paper, a set of six detectors for the above mentioned attributes is designed based on the E-αNet model of neural networks. This model was chosen for its capability to learn hidden activation functions that results in better generalization properties. Experimental set-up and results are presented that show an average 3.5% improvement over a baseline neural network implementation.

Details

Publication typeInproceedings
Published inLecture Notes in Computer Science
PublisherSpringer
> Publications > “Application of EαNets to feature recognitionof articulation manner in knowledge-based automatic speech recognition