Enhanced Running Spectrum Analysis for Robust Speech Recognition Under Adverse Conditions: A Case Study on Japanese Speech

Abstract

In any real environment, noises degrade the performance of Automatic Speech Recognition (ASR) systems. Additionally, in the case of similar pronunciations, it is not easy to realize a high accuracy of recognition. From this point of view, our work envisions an enhanced algorithm processing a speech modulation spectrum, such as Running Spectrum Analysis (RSA). It was also adequately applied to observed speech data. In the envisioned method, a modulation spectrum filtering (MSF) method directly modified the observed cepstral modulation spectrum by a Fourier transform of the cepstral time frequency. The method and experiments carried out for various passbands had favorable results that showed an improvement of about 1-4 % in recognition accuracycompared to conventional methods.

Sound signal analysis in Japanese speech recognition based on deep learning algorithm

“…The literature introduces an improved model of a multi-scale God convolutional network and uses it to ne-tune events [9][10]. The stacked traditional convolutional neural network has the problem of losing representation at a lower level.…”