PLDA Model Research Articles

Multi Kelas Speaker Recognition Menggunakan Deep Learning dengan CN-Celeb Dataset

Speaker recognition has been widely applied in various fields of human life such as Siri from Apple, Cortana from Microsoft, and Voice Assistant by Google. One of the problems when creating speaker recognition is related to the dataset used for the modeling process. The dataset used for creating the speaker recognition model is mostly data that cannot represent real-world conditions. The result is when implemented in the real-world conditions are not optimal. This study develops a speaker recognition model using deep learning (LSTM) with the CN-Celeb dataset. The CN-Celeb dataset is data taken directly from the real world so there is a lot of noise. The hope of using this dataset is that it can represent real world conditions. Model development uses 2 stacked LSTM for multi-class speaker recognition tasks. In addition, this study performs tuning hyperparameters with a grid search method to obtain the most optimal model configuration. The results showed that the EER value of the LSTM model was 10.13% better than the reference baseline paper of 15.52%. In addition, when compared with other studies that also used the CN-Celeb dataset but using different models, it was found that the LSTM model had promising results. From the results of study that has been carried out and also compared with other people's research, it was found that the LSTM model gave promising performance. The LSTM model is compared with the x-vectors, PLDA, TDNN, and transformers models

Speaker-Phrase-Specific Adaptation of PLDA Model for Improved Performance in Text-Dependent Speaker Verification

The i-vector/probabilistic linear discriminant analysis (PLDA) framework has been popularly used in the field of speaker verification for a long time. Lately, the introduction of online i-vectors and its integration with dynamic time warping template matching technique have significantly improved the performance of text-dependent speaker verification system. The PLDA model learns to discriminate among instances of different speaker-phrase classes and also compensates for channel and session variability. However, when exposed to unseen speakers and text, the variability compensation model turns less than optimal, leading to substantial verification error. In this paper, PLDA adaptation, in order to incorporate the idea of speaker-phrase-dependent variability in the ivector/PLDA technique, has been proposed. The adapted model gets specifically tuned to particular speaker-phrase class, leading to a more optimal solution. Two adaptation techniques, namely interpolation and weighted likelihood, have been explored in this work. Experiments have been performed on Part 1 of the RSR2015 database, and relative equal error rate (EER) reductions of up to 58.22% and 45% have been observed for interpolation and weighted likelihood techniques, respectively. The use of speaker-phrase-specific mean and whitening parameters has led to further improvement, resulting in EER reduction of up to 20% relative to that of the adapted models.

Penalized Latent Dirichlet Allocation Model in Single-Cell RNA Sequencing

Single-cell RNA sequencing (scRNA-seq) quantifies RNA transcripts at individual cell level, providing cellular-level resolution of gene expression variation. The scRNA-seq data are counts of RNA transcripts of all genes in species’ genome, which are of very high dimension and contain excessive zero counts. In order to better reduce the data dimension and extract robust and interpretable biological information, we develop a penalized Latent Dirichlet Allocation (pLDA) model for scRNA-seq data. The method is adapted from the generative probabilistic model LDA originated in natural language processing. pLDA models the scRNA-seq data by considering genes as words, cells as documents, and latent biological functions as topics. It imposes a penalty to reflect the characteristics in scRNA-seq that only a small subset of genes are expected to be topic-specific, which increases the robustness of the estimation and interpretability of the results. We apply pLDA to scRNA-seq datasets from both Drop-seq and SMARTer v1 technologies, and demonstrate improved performances in cell-type classification. The topics identified by pLDA are interpretable with biological functions.

SNR-Invariant Multitask Deep Neural Networks for Robust Speaker Verification

A major challenge in speaker verification is to achieve low error rates under noisy environments. We observed that background noise in utterances will not only enlarge the speaker-dependent i -vector clusters but also shift the clusters, with the amount of shift depending on the signal-to-noise ratio (SNR) of the utterances. To overcome this SNR-dependent clustering phenomenon, we propose two deep neural network (DNN) architectures: hierarchical regression DNN (H-RDNN) and multitask DNN (MT-DNN). The H-RDNN is formed by stacking two regression DNNs in which the lower DNN is trained to map noisy i -vectors to their respective speaker-dependent cluster means of clean i -vectors and the upper DNN aims to regularize the outliers that cannot be denoised properly by the lower DNN. The MT-DNN is trained to denoise i -vectors (main task) and classify speakers (auxiliary task). The network leverages the auxiliary task to retain speaker information in the denoised i -vectors. Experimental results suggest that these two DNN architectures together with the PLDA backend significantly outperform the multicondition PLDA model and mixtures of PLDA, and that multitask learning helps to boost verification performance.

Nonparametrically trained PLDA for short duration i-vector speaker verification

The duration of speech segments can significantly impact the performance of text-independent speaker verification systems. In real world applications which require high accuracy on short utterances, the performance of i-vector speaker verification framework degrades significantly considering that i-vectors extracted from short utterances are less reliable (i.e., uncertainty is higher) than those extracted from long utterances. Therefore, to handle duration variability properly, a more realistic approach seems to be required. This study is an extension to our recently proposed nearest neighbor probabilistic linear discriminant analysis (NN-PLDA) which estimates the parameters of PLDA in i-vector speaker verification framework using a nonparametric form rather than maximum likelihood estimation (MLE) obtained by an EM algorithm, and has been shown to provide superior performance. In NN-PLDA, the between-speaker covariance matrix that represents global information about the speaker variability is replaced with a local estimation computed on a nearest neighbor basis for each target speaker. Compared to their parametric counterparts, the nonparametric between- and within-speaker scatter matrices can better exploit the discriminant information in training data and are more adapted to sample distributions. In this paper, we provide further analysis on the proposed nonparametrically trained PLDA as well as introduce a duration variability modeling technique in the estimation of the within-speaker scatter matrix as to compensate for the effect of limited speech data. We evaluate our approach using core–10sec and 10sec–10sec telephone trial conditions of NIST 2010 SRE as well as on the truncated test utterances in extended core condition with duration less than 10 s. We also present the results obtained by the successful incorporation of NN-PLDA on the recent NIST 2016 speaker recognition evaluation. In all experiments, considerable performance improvement is obtained with the proposed technique compared to a generatively trained PLDA model.

Nonlinear I-Vector Transformations for PLDA-Based Speaker Recognition

This paper proposes to estimate parametric nonlinear transformations of i-vectors for speaker recognition systems based on probabilistic linear discriminant analysis PLDA classification. The Gaussian PLDA model assumes that the i-vectors are distributed according to the standard normal distribution. However, it has been shown that the i-vectors are better modeled, for example, by Heavy-Tailed distributions, and that significant improvement of the classification performance can be obtained by whitening and length normalizing the i-vectors. In this paper, we propose to transform the i-vectors so that their distribution becomes more suitable to discriminate speakers using the PLDA model. This is performed by means of a sequence of affine and nonlinear transformations whose parameters are obtained by maximum likelihood estimation on the development set. Another contribution of this paper is the reduction of the mismatch between the development and evaluation i-vector length distributions by means of a scaling factor tuned for the estimated i-vector distribution, rather than by means of a blind length normalization. Relative improvement between 7% and 14% of the detection cost function was obtained with the proposed technique on the NIST SRE-2010 and SRE-2012 evaluation datasets, using both the traditional GMM/UBM and the hybrid DNN/GMM-based systems.

Direct Optimization of the Detection Cost for I-Vector-Based Spoken Language Recognition

We explore a method to boost discriminative capabilities of probabilistic linear discriminant analysis PLDA model without losing its generative advantages. We show a sequential projection and training steps leading to a classifier that operates in the original i-vector space but is discriminatively trained in a low-dimensional PLDA latent subspace. We use extended Baum-Welch technique to optimize the model with respect to two objective functions for discriminative training. One of them is the well-known maximum mutual information objective, while the other one is a new objective that we propose to approximate the language detection cost. We evaluate the performance on NIST language recognition evaluation LRE 2015 and our development dataset comprised of the utterances from previous LREs. We improve the detection cost by 10% and 6% relative compared to our fine-tuned generative and discriminative baselines, and by 10% over the best of our previously reported results. The proposed approximation method of the cost function and PLDA subspace training are applicable for a broad range of tasks.

A Large-Scale Open-Source Acoustic Simulator for Speaker Recognition

The state-of-the-art speaker-recognition systems suffer from significant performance loss on degraded speech conditions and acoustic mismatch between enrolment and test phases. Past international evaluation campaigns, such as the NIST speaker recognition evaluation (SRE), have partly addressed these challenges in some evaluation conditions. This work aims at further assessing and compensating for the effect of a wide variety of speech-degradation processes on speaker-recognition performance. We present an open-source simulator generating degraded telephone, VoIP, and interview-speech recordings using a comprehensive list of narrow-band, wide-band, and audio codecs, together with a database of over 60 h of environmental noise recordings and over 100 impulse responses collected from publicly available data. We provide speaker-verification results obtained with an $i$ -vector-based system using either a clean or degraded PLDA back-end on a NIST SRE subset of data corrupted by the proposed simulator. While error rates increase considerably under degraded speech conditions, large relative equal error rate (EER) reductions were observed when using a PLDA model trained with a large number of degraded sessions per speaker.

Mixture of PLDA for Noise Robust I-Vector Speaker Verification

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 i-vector 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 .

Large-Scale Training of Pairwise Support Vector Machines for Speaker Recognition

State-of-the-art systems for text-independent speaker recognition use as their features a compact representation of a speaker utterance, known as i-vector. We recently presented an efficient approach for training a Pairwise Support Vector Machine (PSVM) with a suitable kernel for i-vector pairs for a quite large speaker recognition task. Rather than estimating an SVM model per speaker, according to the versus all discriminative paradigm, the PSVM approach classifies a trial, consisting of a pair of i-vectors, as belonging or not to the same speaker class. Training a PSVM with large amount of data, however, is a memory and computational expensive task, because the number of training pairs grows quadratically with the number of training i-vectors. This paper demonstrates that a very small subset of the training pairs is necessary to train the original PSVM model, and proposes two approaches that allow discarding most of the training pairs that are not essential, without harming the accuracy of the model. This allows dramatically reducing the memory and computational resources needed for training, which becomes feasible with large datasets including many speakers. We have assessed these approaches on the extended core conditions of the NIST 2012 Speaker Recognition Evaluation. Our results show that the accuracy of the PSVM trained with a sufficient number of speakers is 10%-30% better compared to the one obtained by a PLDA model, depending on the testing conditions. Since the PSVM accuracy increases with the training set size, but PSVM training does not scale well for large numbers of speakers, our selection techniques become relevant for training accurate discriminative classifiers.

On the use of i–vector posterior distributions in Probabilistic Linear Discriminant Analysis

The i-vector extraction process is affected by several factors such as the noise level, the acoustic content of the observed features, the channel mismatch between the training conditions and the test data, and the duration of the analyzed speech segment. These factors influence both the i-vector estimate and its uncertainty, represented by the i-vector posterior covariance. This paper presents a new PLDA model that, unlike the standard one, exploits the intrinsic i-vector uncertainty. Since the recognition accuracy is known to decrease for short speech segments, and their length is one of the main factors affecting the i-vector covariance, we designed a set of experiments aiming at comparing the standard and the new PLDA models on short speech cuts of variable duration, randomly extracted from the conversations included in the NIST SRE 2010 extended dataset, both from interviews and telephone conversations. Our results on NIST SRE 2010 evaluation data show that in different conditions the new model outperforms the standard PLDA by more than 10% relative when tested on short segments with duration mismatches, and is able to keep the accuracy of the standard model for long enough speaker segments. This technique has also been successfully tested in the NIST SRE 2012 evaluation.