Frame level likelihood normalization for text-independent speaker identification using Gaussian mixture models
Konstantin Markov, Seiichi Nakagawa · 2002
Proposes a new speaker identification system, where the likelihood normalization technique, which is widely used for speaker verification, is introduced. In the new system, which is based on Gaussian mixture models, every frame of the test utterance is input to all the reference models in parallel. In this procedure, for each frame, likelihoods from all the models are available, and hence they can be normalized at every frame. A special kind of likelihood normalization, called the 'weighting models rank', is also proposed. Experiments were performed using two databases-TIMIT and NTT. Evaluation results clearly show that the frame-level likelihood normalization technique is superior to the standard accumulated likelihood approach.