Modeling acoustic-phonetic detail in an HMM-based large vocabulary speech recognizer | IEEE Conference Publication | IEEE Xplore