In real-world environments, noisy utterances with variable noise levels are recorded and then converted to i-vectors for cosine distance or PLDA scoring. This paper investigates the effect of noise-level variability on i-vectors. It demonstrates that noise-level variability causes the i-vectors to shift, causing the noise contaminated i-vectors to form clusters in the ivector space. It also demonstrates that optimal subspaces for discriminating speakers are noise-level dependent. Based on these observations, this paper proposes using signal-to-noise ratio (SNR) of utterances as guidance for training mixture of PLDA models. To maximize the coordination among the PLDA models, mixtures of PLDA models are trained simultaneously via an EM algorithm using the utterances contaminated with noise at various levels. For scoring, given a test i-vector, the marginal likelihoods from individual PLDA models are linearly combined by the posterior probabilities of the test utterance's SNR. Verification scores are the ratio of the marginal likelihoods. Results based on NIST 2012 SRE suggest that the SNR-dependent mixture of PLDA is not only suitable for the situations where the test utterances exhibit a wide range of SNR, but also beneficial for the test utterances with unknown SNR distribution. Supplementary materials containing full derivations of the EM algorithms and scoring functions can be found in http://bioinfo.eie.polyu.edu.hk/mPLDA/SuppMaterials.pdf.

The i-vector representation and probabilistic linear discriminant analysis (PLDA) have shown state-of-the-art performance in many speaker verification systems. However, in real-world en vironments, additive and convolutive noise cause mismatches between training and recognition conditions, degrading the per formance. In this paper, a fusion system that combines a multi condition PLDA model and a mixture of SNR-dependent PLDA models is proposed to make the verification system noise ro bust. The SNR of test utterances is used to determine the best SNR-dependent PLDA model to score against the target speaker's i-vectors. The performance of the fusion system is demonstrated on NIST 2012 SRE. Results show that the SNR dependent PLDA models can reduce EER and that the fusion system is more robust than the conventional i-vector/PLDA sys tems under noisy conditions. It is also found that the SNR dependent PLDA models are insensitive to Z-norm parameters.

While i-vectors with probabilistic linear discriminant analysis (PLDA) can achieve state-of-the-art performance in speaker verification, the mismatch caused by acoustic noise remains a key factor affecting system performance. In this paper, a fusion system that combines a multi-condition SNR-independent PLDA model and a mixture of SNR-dependent PLDA models is proposed to make speaker verification systems more noise robust. First, the whole range of SNR that a verification system is expected to operate is divided into several narrow ranges. Then, a set of SNR-dependent PLDA models, one for each narrow SNR range, are trained. During verification, the SNR of the test utterance is used to determine which of the SNR-dependent PLDA models is used for scoring. To further enhance performance, the SNR-dependent and SNRindependent models are fused using linear and logistic regression fusion. The performance of the fusion system and the SNR-dependent system is evaluated on the NIST 2012 SRE for both noisy and clean conditions. Results show that a mixture of SNR-dependent PLDA models perform better in both clean and noisy conditions. It was also found that the fusion system is more robust than the conventional i-vector/PLDA systems under noisy conditions. KeywordsSpeaker verification • i-vectors • probabilistic LDA • NIST 2012 SRE • noise robustness • fusion This is the Pre-Published Version.