HIDDEN CONDITIONAL RANDOM FIELD MODELS FOR PHONETIC CLASSIFICATION AND SPEECH RECOGNITION

摘要

Speech recognition method and apparatus used to train and Hidden Conditional Random Field model for voice classification is provided . Hidden conditional random field model is to use the features , at least one of the features is based on the hidden state in the audio unit . Values for the features are determined from a segment of speech , then this value is used to identify a phonetic unit for the speech segment .