Discriminative feature weighting for HMM-based continuous speech recognizers

The Discriminative Feature Extraction (DFE) method provides an appropriate formalism for the design of the front-end feature extraction module in pattern classification systems. In the recent years, this formalism has been successfully applied to different speech recognition problems, like classification of vowels, classification of phonemes or isolated word recognition. The DFE formalism can be applied to weight the contribution of the components in the feature vector. This variant of DFE, that we call Discriminative Feature Weighting (DFW), improves the pattern classification systems by enhancing those components more relevant for the discrimination among the different classes. This paper is dedicated to the application of the DFW formalism to Continuous Speech Recognizers (CSR) based on Hidden Markov Models (HMMs). Two different types of HMM-based speech recognizers are considered: recognizers based on Discrete-HMMs (DHMMs) (for which the acoustic evaluation is based on an Euclidean distance measure) and Semi-Continuous-HMMs (SCHMMs) (for which the acoustic evaluation is performed making use of a mixture of multivariated Gaussians). We report how the components can be weighted and how the weights can be discriminatively trained and applied to the speech recognizers. We present recognition results for several continuous speech recognition tasks. The experimental results show the utility of DFW for HMM-based continuous speech recognizers. Die Methode der diskriminativen Merkmalextraktion (Discriminative Feature Extraction, DFE) stellt einen für den Entwurf des Eingangsmerkmalextraktionsmoduls in Musterklassifizierungssystemen geeigneten Algorithmus zur Verfügung. In vergangenen Jahren wurde dieser Algorithmus erfolgreich auf verschiedene Spracherkennungsprobleme wie die Klassifizierung von Selbstlauten, Phonemklassifizierung, oder die Erkennung isolierter Wörter angewandt. Der DFE Algorithmus kann zur Gewichtung des Beitrags der Komponenten des Merkmalvektors verwendet werden. Diese Variante der DFE, die wir diskriminative Merkmalgewichtung (Discriminative Feature Weighting, DFW) nennen, verbessert Musterklassifizierungssysteme, indem sie jene Komponenten verstärkt, die mehr Relevanz bei der Unterscheidung verschiedener Klassen haben. Diese Veröffentlichung widmet sich der Anwendung der diskriminativen Merkmalgewichtung auf Systeme zur Erkennung kontinuierlicher Sprache, die auf “Hidden Markov Models” (HMM) beruhen. Es werden zwei verschiedene Arten von H