Toward noise-tolerant acoustic models

State-of-the-art acoustic models, relying on hidden Markov models (HMMs), are heavily noise-sensitive: recognition performance drops whenever a significant difference in acoustic conditions holds between the training and the test environments. Practical, yet partial, attempts to tackle the problem are usually based on noise reduction via spectral subtraction, blind source separation, parameter adaptation/normalization, HMM retraining, etc. But the relevance of developing acoustic models that are intrinsically robust has to be stressed. Robustness to noise is related to the generalization capabilities of the model. Artificial neural networks (ANNs) appear to be a promising alternative, but they historically failed as a general paradigm for speech recognition [1]. This paper faces the problem by (i) investigating the recognition performance of the ANN/HMM hybrid proposed by the authors [2] over tasks with noisy signals, and (ii) proposing an explicit “soft” weight grouping technique, capable to improve its robustness. Experiments over noisy speakerindependent connected-digits strings are presented. In particular, results on the VODIS II/SpeechDatCar database, collected in a real car environment, show the dramatic gain over the standard HMM, as well as over Bourlard and Morgan’s hybrid.