Speech recognition method decision tree construction method and apparatus for speech recognition

摘要

The present invention relates to a speech recognition method, a decision tree constructing method for speech recognition, and an apparatus for speech recognition, wherein a normal phoneme not including a pronunciation error and an error phoneme including a pronunciation error are distinguished from each other, A second decision tree having fewer branches than the normal phoneme is constructed by grouping the decision trees based on a higher threshold value than the normal phoneme in the case of error phoneme, A voice recognition system for performing voice recognition and error detection on a user's voice by using a decision tree to prevent an increase in variance and thereby detecting an error pronunciation using a decision tree for improving an error pronunciation detection capability; do.