Combining a group of classifiers and therefore improving the overall classification performance is a young and promising direction in Large Vocabulary Continuous Speech Recognition (LVCSR). Previous works on acoustic ...