Missing data methods attempt to improve robust speech recognition by distinguishing between reliable and unreliable data in the time-frequency domain. Such methods require a binar...
This paper describes a series of experiments to test the hypothesis that the parallel application of multiple NLP tools and the integration of their results improves the correctne...
The problem of decentralized power allocation for competitive rate maximization in a frequency-selective Gaussian interference channel is considered. In the absence of perfect kno...
Amod J. G. Anandkumar, Animashree Anandkumar, Sang...
We present an approach to music identification based on weighted finite-state transducers and Gaussian mixture models, inspired by techniques used in large-vocabulary speech recogn...
This paper considers the problem of weak signal detection in the presence of navigation data bits for Global Navigation Satellite System (GNSS) receivers. Typically, a set of part...