A Robust Frame-based Nonlinear Prediction System for Automatic Speech Coding

TL;DR

This paper introduces a neural-based speech coding system that uses a two-stage training process to accurately compress and reconstruct speech signals, demonstrating robustness across speakers and phonemes in both time and DCT domains.

Contribution

It presents a novel frame-based nonlinear predictive coding method utilizing neural networks with a two-stage training process for improved speech signal reconstruction.

Findings

Effective reconstruction of phonemes with good accuracy.

DCT domain training yields better performance for energy-rich frames.

System demonstrates robustness across different speakers and utterances.

Abstract

In this paper, we propose a neural-based coding scheme in which an artificial neural network is exploited to automatically compress and decompress speech signals by a trainable approach. Having a two-stage training phase, the system can be fully specified to each speech frame and have robust performance across different speakers and wide range of spoken utterances. Indeed, Frame-based nonlinear predictive coding (FNPC) would code a frame in the procedure of training to predict the frame samples. The motivating objective is to analyze the system behavior in regenerating not only the envelope of spectra, but also the spectra phase. This scheme has been evaluated in time and discrete cosine transform (DCT) domains and the output of predicted phonemes show the potentiality of the FNPC to reconstruct complicated signals. The experiments were conducted on three voiced plosive phonemes, b/d/g/ in time and DCT domains versus the number of neurons in the hidden layer. Experiments approve the FNPC capability as an automatic coding system by which /b/d/g/ phonemes have been reproduced with a good accuracy. Evaluations revealed that the performance of FNPC system, trained to predict DCT coefficients is more desirable, particularly for frames with the wider distribution of energy, compared to time samples.