Combination of speech features using smoothed heteroscedastic linear discriminant analysis
نویسنده
چکیده
Feature combination techniques based on PCA, LDA and HLDA are compared in experiments where limited amount of training data is available. Success with feature combination can be quite dependent on proper estimation of statistics required by the used technique. Insufficiency of training data is, therefore, an important problem, which has to be taken in to account in our experiments. Besides of some standard approaches increasing robustness of statistic estimation, methods based on combination of LDA and HLDA are proposed. An improved recognition performance obtained using these methods is demonstrated in experiments.
منابع مشابه
Pitch adaptive features for LVCSR
We have investigated the use of a pitch adaptive spectral representation on large vocabulary speech recognition, in conjunction with speaker normalisation techniques. We have compared the effect of a smoothed spectrogram to the pitch adaptive spectral analysis by decoupling these two components of STRAIGHT. Experiments performed on a large vocabulary meeting speech recognition task highlight th...
متن کاملLong Span Features and Minimum Phoneme Error Heteroscedastic Linear Discriminant Analysis
In this paper we explore the effect of long-span features, resulting from concatenating multiple speech frames and projecting the resulting vector onto a subspace using Linear Discriminant Analysis (LDA) techniques. We show that LDA is not always effective in selecting the optimal combination of long-span features, and introduce a discriminative feature analysis method that seeks to minimize ph...
متن کاملLinear Discriminant Analysis Using a Generalized Mean of Class Covariances and Its Application to Speech Recognition
To precisely model the time dependency of features is one of the important issues for speech recognition. Segmental unit input HMM with a dimensionality reduction method has been widely used to address this issue. Linear discriminant analysis (LDA) and heteroscedastic extensions, e.g., heteroscedastic linear discriminant analysis (HLDA) or heteroscedastic discriminant analysis (HDA), are popula...
متن کاملReview on Heteroscedastic Discriminant Analysis
Discriminant feature spaces are attractive way to improve the word error rate performance of the speech recognition systems. Heteroscedastic discriminant analysis (HDA) is a generalized method for the feature space transformation that does not impose the equa l w i th in c l a s s cova r i ance assumptions required by the standard linear discriminant analysis (LDA). It will be shown that the co...
متن کاملRecent Advances in Broadcast News Transcription
This paper describes recent advances in the CU-HTK Broadcast News English (BN-E) transcription system and its performance in the DARPA/NIST Rich Transcription 2003 Speech-to-Text (RT03) evaluation. Heteroscedastic linear discriminant analysis (HLDA) and discriminative training, which were previously developed in the context of the recognition of conversational telephone speech, have been succes...
متن کامل