fMPE: discriminatively trained features for speech recognition

• Published in: Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005 Vol. 1; pp. I/961 - I/964 Vol. 1
• Main Authors: Povey, D., Kingsbury, B., Mangu, L., Saon, G., Soltau, H., Zweig, G.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 2005
• Subjects: Error analysis; Gaussian processes; Hidden Markov models; Robustness; Speech recognition; Statistics; Testing; Training data
• ISBN: 9780780388741; 0780388747
• ISSN: 1520-6149; 2379-190X
• DOI: 10.1109/ICASSP.2005.1415275

MPE (minimum phone error) is a previously introduced technique for discriminative training of HMM parameters. fMPE applies the same objective function to the features, transforming the data with a kernel-like method and training millions of parameters, comparable to the size of the acoustic model. Despite the large number of parameters, fMPE is robust to over-training. The method is to train a matrix projecting from posteriors of Gaussians to a normal size feature space, and then to add the projected features to normal features such as PLP. The matrix is trained from a zero start using a linear method. Sparsity of posteriors ensures speed in both training and test time. The technique gives similar improvements to MPE (around 10% relative). MPE on top of fMPE results in error rates up to 6.5% relative better than MPE alone, or more if multiple layers of transform are trained.