Missing Reliability Correction In Modality Information Integration For Robust Speaker Identification

In the emerging biometrics technology, speaker identification in real environment is one of the key issues for enhancing the density of human computer interaction. In this paper, we propose an optimizing factor through a fuzzy membership function for correcting the reliability in different modalities reliability measure in a bimodal fusion process for speaker identification. In the bimodal speaker identification system, we have also applied our proposed extended modified convection reliability function to account optimal reliability simultaneously for audio and visual information integration. For creating mismatch in between train and test stage, babble noises and artificial illumination have been added to test speeches and lip images, respectively. Local PCA have been applied at features level at both classifiers system for reducing the dimension of features vector at different level of signal distortion. We have applied particle swarm optimization (PSO) for optimizing the proposed fuzzy membership function based optimizing factor and modified convection function's optimizing parameters. The speaker identification experiments have been performed using VidTimit database. Experimental results show that our proposed method enhanced the identification accuracy in comparison with the baseline system thus demonstrated the validation of the proposed approach and most notably maintains the consistency of the integration process.