High resolution signal reconstruction

TLDR: A framework for speech enhancement and robust speech recognition that exploits the harmonic structure of speech and achieves substantial gains in signal-to-noise ratio (SNR) of enhanced speech as well as considerable gains in accuracy of automatic speech recognition in very noisy conditions.

Abstract: We present a framework for speech enhancement and robust speech recognition that exploits the harmonic structure of speech. We achieve substantial gains in signal-to-noise ratio (SNR) of enhanced speech as well as considerable gains in accuracy of automatic speech recognition in very noisy conditions. The method exploits the harmonic structure of speech by employing a high frequency resolution speech model in the log-spectrum domain and reconstructs the signal from the estimated posteriors of the clean signal and the phases from the original noisy signal. We achieve a gain in SNR of 8.38 dB for enhancement of speech at 0 dB. We also present recognition results on the Aurora 2 data-set. At 0 dB SNR, we achieve a reduction of relative word error rate of 43.75% over the baseline, and 15.90% over the equivalent low-resolution algorithm.

Figures

Fig. 5. Change in absolute Word Accuracy of HighResolution Signal Reconstruction using Gender Dependent models and Low-Resolution Algonquin compared to the Aurora Multi-condition Baseline for the Car noise condition.

Fig. 2. Spectrogram of clean speech. The words ’TWO FIVE’ are being spoken.