13th Annual Conference of the International Speech Communication Association

Portland, OR, USA
September 9-13, 2012

Continuous Digit Recognition in Noise: Reservoirs Can Do an Excellent Job!

Azarakhsh Jalalvand, Fabian Triefenbach, Jean-Pierre Martens

Ghent University - IBBT, ELIS Multimedia Lab, Ghent, Belgium

In this paper a formerly proposed continuous digit recognition system based on Reservoir Computing (RC) is improved in two respects: (1) the single reservoir is substituted by a stack of reservoirs, and (2) the straightforward mapping of reservoir outputs to state likelihoods is replaced by a trained non-parametric mapping. Furthermore, it is shown that a reservoir-based method can improve a model trained on clean speech to work better in a noisy condition from which it has a number of unknown digit string recordings available. The first two improvements have lead to a system that outperforms a HMM-based system with the same noise robust features as input. The model adaptation offers a significant supplementary gain when the noise level is not too high.

Index Terms: Reservoir Computing, Acoustic Modeling, Model Adaptation, Noise Robustness

Full Paper

Bibliographic reference.  Jalalvand, Azarakhsh / Triefenbach, Fabian / Martens, Jean-Pierre (2012): "Continuous digit recognition in noise: reservoirs can do an excellent job!", In INTERSPEECH-2012, 1804-1807.