Learning improved linear transforms for speech recognition
Acoustics, Speech and Signal Processing（2012）
摘要
This paper explores a novel large margin approach to learning a linear transform for dimensionality reduction in speech recognition. The method assumes a trained Gaussian mixture model for each class to be discriminated and trains a dimensionality-reducing linear transform with respect to the fixed model, optimizing a hinge loss on the difference between the distance to the nearest in- and out-of-class Gaussians using stochastic gradient descent. Results are presented showing that the learnt transform improves state classification for individual frames and reduces word error rate compared to Linear Discriminant Analysis (LDA) in a large vocabulary speech recognition problem even after discriminative training.
更多查看译文
关键词
Gaussian processes,gradient methods,speech recognition,stochastic processes,transforms,vocabulary,LDA,dimensionality-reducing linear transform,discriminative training,hinge loss optimization,individual frame classification,learning improved linear transform,linear discriminant analysis,nearest in-of-class Gaussian,nearest out-of-class Gaussian,stochastic gradient descent,trained Gaussian mixture model,vocabulary speech recognition problem,word error rate reduction,LDA,Linear discriminant analysis,margin Mahalanobis distance,speech feature transformation,stochastic gradient descent
AI 理解论文
溯源树
样例
生成溯源树，研究论文发展脉络
Chat Paper
正在生成论文摘要