Technical Program

Paper Detail

Paper:	SLP-P11.1
Session:	Front-end For Robust Speech Recognition
Time:	Wednesday, May 17, 16:30 - 18:30
Presentation:	Poster
Topic:	Speech and Spoken Language Processing: Feature-based Robust Speech Recognition (e.g., noise, etc)
Title:	Application of Minimum Statistics and Minima Controlled Recursive Averaging Methods to Estimate a Cepstral Noise Model for Robust ASR
Authors:	Veronique Stouten, Hugo Van hamme, Patrick Wambacq, Katholieke Universiteit Leuven, Belgium
Abstract:	Many compensation techniques, both in the model and feature domain, require an estimate of the noise statistics to compensate for the clean speech degradation in adverse environments. We explore how two spectral noise estimation approaches can be applied in the context of model-based feature enhancement. The minimum statistics method and the improved minima controlled recursive averaging method are used to estimate the noise power spectrum based only on the noisy speech. The noise mean and variance estimates are non-linearly transformed to the cepstral domain and used in the Gaussian noise model of MBFE. Then, an MMSE-reestimation of the noise mean is done, based on a clean speech model. MBFE combined with MS or IMCRA achieves an accuracy on the Aurora2 recognition task that is comparable to MBFE with a fixed noise model trained on noise, while the reestimation of the noise mean significantly improves its performance.