Acoustic model training using committee-based active and semi-supervised learning for speech recognition | IEEE Conference Publication | IEEE Xplore