Improved acoustic modeling for speaker independent large vocabulary continuous speech recognition | IEEE Conference Publication | IEEE Xplore