Preliminary Study on SSCF-derived Polar Coordinate for ASR

TL;DR

This study explores the use of SSCF-derived polar coordinates to better characterize speech signals for automatic speech recognition, showing improved accuracy over traditional angles, especially with derivatives, but limited gender independence.

Contribution

It introduces the use of polar coordinates derived from SSCF for speech signal representation, demonstrating their superiority over angles in ASR accuracy.

Findings

Polar coordinates outperform angles in mixed and cross-gender recognition.

Using derivatives ($ abla$, $ abla abla$) further improves accuracy.

Gender independence remains limited compared to MFCCs.

Abstract

The transition angles are defined to describe the vowel-to-vowel transitions in the acoustic space of the Spectral Subband Centroids, and the findings show that they are similar among speakers and speaking rates. In this paper, we propose to investigate the usage of polar coordinates in favor of angles to describe a speech signal by characterizing its acoustic trajectory and using them in Automatic Speech Recognition. According to the experimental results evaluated on the BRAF100 dataset, the polar coordinates achieved significantly higher accuracy than the angles in the mixed and cross-gender speech recognitions, demonstrating that these representations are superior at defining the acoustic trajectory of the speech signal. Furthermore, the accuracy was significantly improved when they were utilized with their first and second-order derivatives ($\Delta$, $\Delta$$\Delta$), especially in cross-female recognition. However, the results showed they were not much more gender-independent than the conventional Mel-frequency Cepstral Coefficients (MFCCs).