Feature Combination Using Multiple Spectral Cues for Robust Speech Recognition in Mobile Communications

This paper investigates a new front-end processing that aims at improving the performance of speech recognition in noisy mobile environments. This approach combines features based on conventional Mel-cepstral coefficients (MFCCs) and line spectral frequencies (LSFs) to constitute robust multivariate feature vectors. The proposed front-end constitutes an alternative to the DSRXAFE (XAFE: extended audio front-end) available in GSM mobile communications. Our results showed that for highly noisy speech, using the paradigm that combines LSF with MFCCs, leads to a significant improvement in recognition accuracy on the Aurora 2 task.