arXiv subjects

Hakan Erdogan

Publications and source records attributed to Hakan Erdogan.

34 records · Page 2Linked to original sources

arXiv2019

Universal Sound Separation

Recent deep learning approaches have achieved impressive performance on speech enhancement and separation tasks. However, these approaches have not been investigated for separating mixtures of arbitrary sounds of different types, a task we refer to as universal sound separation, and it is unknown how performance on speech tasks carries over to non-speech tasks. To study this question, we develop a dataset of mixtures containing arbitrary sounds, and use it to investigate the space of mask-based separation architectures, varying both the overall network architecture and the framewise analysis-synthesis basis for signal transformations. These network architectures include convolutional long short-term memory networks and time-dilated convolution stacks inspired by the recent success of time-domain enhancement networks like ConvTasNet. For the latter architecture, we also propose novel modifications that further improve separation performance. In terms of the framewise analysis-synthesis basis, we explore both a short-time Fourier transform (STFT) and a learnable basis, as used in ConvTasNet. For both of these bases, we also examine the effect of window size. In particular, for STFTs, we find that longer windows (25-50 ms) work best for speech/non-speech separation, while shorter windows (2.5 ms) work best for arbitrary sounds. For learnable bases, shorter windows (2.5 ms) work best on all tasks. Surprisingly, for universal sound separation, STFTs outperform learnable bases. Our best methods produce an improvement in scale-invariant signal-to-distortion ratio of over 13 dB for speech/non-speech separation and close to 10 dB for universal sound separation.

cs.SD↗

arXiv2019

Low-Latency Speaker-Independent Continuous Speech Separation

Speaker independent continuous speech separation (SI-CSS) is a task of converting a continuous audio stream, which may contain overlapping voices of unknown speakers, into a fixed number of continuous signals each of which contains no overlapping speech segment. A separated, or cleaned, version of each utterance is generated from one of SI-CSS's output channels nondeterministically without being split up and distributed to multiple channels. A typical application scenario is transcribing multi-party conversations, such as meetings, recorded with microphone arrays. The output signals can be simply sent to a speech recognition engine because they do not include speech overlaps. The previous SI-CSS method uses a neural network trained with permutation invariant training and a data-driven beamformer and thus requires much processing latency. This paper proposes a low-latency SI-CSS method whose performance is comparable to that of the previous method in a microphone array-based meeting transcription task.This is achieved (1) by using a new speech separation network architecture combined with a double buffering scheme and (2) by performing enhancement with a set of fixed beamformers followed by a neural post-filter.

eess.AS↗

arXiv2018

SDR - half-baked or well done?

In speech enhancement and source separation, signal-to-noise ratio is a ubiquitous objective measure of denoising/separation quality. A decade ago, the BSS_eval toolkit was developed to give researchers worldwide a way to evaluate the quality of their algorithms in a simple, fair, and hopefully insightful way: it attempted to account for channel variations, and to not only evaluate the total distortion in the estimated signal but also split it in terms of various factors such as remaining interference, newly added artifacts, and channel errors. In recent years, hundreds of papers have been relying on this toolkit to evaluate their proposed methods and compare them to previous works, often arguing that differences on the order of 0.1 dB proved the effectiveness of a method over others. We argue here that the signal-to-distortion ratio (SDR) implemented in the BSS_eval toolkit has generally been improperly used and abused, especially in the case of single-channel separation, resulting in misleading results. We propose to use a slightly modified definition, resulting in a simpler, more robust measure, called scale-invariant SDR (SI-SDR). We present various examples of critical failure of the original SDR that SI-SDR overcomes.

cs.SD↗

arXiv2018

Recognizing Overlapped Speech in Meetings: A Multichannel Separation Approach Using Neural Networks

The goal of this work is to develop a meeting transcription system that can recognize speech even when utterances of different speakers are overlapped. While speech overlaps have been regarded as a major obstacle in accurately transcribing meetings, a traditional beamformer with a single output has been exclusively used because previously proposed speech separation techniques have critical constraints for application to real meetings. This paper proposes a new signal processing module, called an unmixing transducer, and describes its implementation using a windowed BLSTM. The unmixing transducer has a fixed number, say J, of output channels, where J may be different from the number of meeting attendees, and transforms an input multi-channel acoustic signal into J time-synchronous audio streams. Each utterance in the meeting is separated and emitted from one of the output channels. Then, each output signal can be simply fed to a speech recognition back-end for segmentation and transcription. Our meeting transcription system using the unmixing transducer outperforms a system based on a state-of-the-art neural mask-based beamformer by 10.8%. Significant improvements are observed in overlapped segments. To the best of our knowledge, this is the first report that applies overlapped speech recognition to unconstrained real meeting audio.

eess.AS↗

arXiv2017

Deep Long Short-Term Memory Adaptive Beamforming Networks For Multichannel Robust Speech Recognition

Far-field speech recognition in noisy and reverberant conditions remains a challenging problem despite recent deep learning breakthroughs. This problem is commonly addressed by acquiring a speech signal from multiple microphones and performing beamforming over them. In this paper, we propose to use a recurrent neural network with long short-term memory (LSTM) architecture to adaptively estimate real-time beamforming filter coefficients to cope with non-stationary environmental noise and dynamic nature of source and microphones positions which results in a set of timevarying room impulse responses. The LSTM adaptive beamformer is jointly trained with a deep LSTM acoustic model to predict senone labels. Further, we use hidden units in the deep LSTM acoustic model to assist in predicting the beamforming filter coefficients. The proposed system achieves 7.97% absolute gain over baseline systems with no beamforming on CHiME-3 real evaluation set.

eess.AS↗

arXiv2017

PLDA-Based Diarization of Telephone Conversations

This paper investigates the application of the probabilistic linear discriminant analysis (PLDA) to speaker diarization of telephone conversations. We introduce using a variational Bayes (VB) approach for inference under a PLDA model for modeling segmental i-vectors in speaker diarization. Deterministic annealing (DA) algorithm is imposed in order to avoid local optimal solutions in VB iterations. We compare our proposed system with a well-known system that applies k-means clustering on principal component analysis (PCA) coefficients of segmental i-vectors. We used summed channel telephone data from the National Institute of Standards and Technology (NIST) 2008 Speaker Recognition Evaluation (SRE) as the test set in order to evaluate the performance of the proposed system. We achieve about 20% relative improvement in Diarization Error Rate (DER) compared to the baseline system.

eess.AS↗

arXiv2015

Comments On "Multipath Matching Pursuit" by Kwon, Wang and Shim

Straightforward combination of tree search with matching pursuits, which was suggested in 2001 by Cotter and Rao, and then later developed by some other authors, has been revisited recently as multipath matching pursuit (MMP). In this comment, we would like to point out some major issues regarding this publication. First, the idea behind MMP is not novel, and the related literature has not been properly referenced. MMP has not been compared to closely related algorithms such as A* orthogonal matching pursuit (A*OMP). The theoretical analyses do ignore the pruning strategies applied by the authors in practice. All these issues have the potential to mislead the reader and lead to misinterpretation of the results. With this short paper, we intend to clarify the relation of MMP to existing literature in the area and compare its performance with A*OMP.

cs.IT↗

arXiv2015

Improving A*OMP: Theoretical and Empirical Analyses With a Novel Dynamic Cost Model

Best-first search has been recently utilized for compressed sensing (CS) by the A* orthogonal matching pursuit (A*OMP) algorithm. In this work, we concentrate on theoretical and empirical analyses of A*OMP. We present a restricted isometry property (RIP) based general condition for exact recovery of sparse signals via A*OMP. In addition, we develop online guarantees which promise improved recovery performance with the residue-based termination instead of the sparsity-based one. We demonstrate the recovery capabilities of A*OMP with extensive recovery simulations using the adaptive-multiplicative (AMul) cost model, which effectively compensates for the path length differences in the search tree. The presented results, involving phase transitions for different nonzero element distributions as well as recovery rates and average error, reveal not only the superior recovery accuracy of A*OMP, but also the improvements with the residue-based termination and the AMul cost model. Comparison of the run times indicate the speed up by the AMul cost model. We also demonstrate a hybrid of OMP and A?OMP to accelerate the search further. Finally, we run A*OMP on a sparse image to illustrate its recovery performance for more realistic coefcient distributions.

cs.IT↗

arXiv2014

THRIVE: Threshold Homomorphic encryption based secure and privacy preserving bIometric VErification system

In this paper, we propose a new biometric verification and template protection system which we call the THRIVE system. The system includes novel enrollment and authentication protocols based on threshold homomorphic cryptosystem where the private key is shared between a user and the verifier. In the THRIVE system, only encrypted binary biometric templates are stored in the database and verification is performed via homomorphically randomized templates, thus, original templates are never revealed during the authentication stage. The THRIVE system is designed for the malicious model where the cheating party may arbitrarily deviate from the protocol specification. Since threshold homomorphic encryption scheme is used, a malicious database owner cannot perform decryption on encrypted templates of the users in the database. Therefore, security of the THRIVE system is enhanced using a two-factor authentication scheme involving the user's private key and the biometric data. We prove security and privacy preservation capability of the proposed system in the simulation-based model with no assumption. The proposed system is suitable for applications where the user does not want to reveal her biometrics to the verifier in plain form but she needs to proof her physical presence by using biometrics. The system can be used with any biometric modality and biometric feature extraction scheme whose output templates can be binarized. The overall connection time for the proposed THRIVE system is estimated to be 336 ms on average for 256-bit biohash vectors on a desktop PC running with quad-core 3.2 GHz CPUs at 10 Mbit/s up/down link connection speed. Consequently, the proposed system can be efficiently used in real life applications.

cs.CR↗

arXiv2014

Confidence-Based Dynamic Classifier Combination For Mean-Shift Tracking

We introduce a novel tracking technique which uses dynamic confidence-based fusion of two different information sources for robust and efficient tracking of visual objects. Mean-shift tracking is a popular and well known method used in object tracking problems. Originally, the algorithm uses a similarity measure which is optimized by shifting a search area to the center of a generated weight image to track objects. Recent improvements on the original mean-shift algorithm involves using a classifier that differentiates the object from its surroundings. We adopt this classifier-based approach and propose an application of a classifier fusion technique within this classifier-based context in this work. We use two different classifiers, where one comes from a background modeling method, to generate the weight image and we calculate contributions of the classifiers dynamically using their confidences to generate a final weight image to be used in tracking. The contributions of the classifiers are calculated by using correlations between histograms of their weight images and histogram of a defined ideal weight image in the previous frame. We show with experiments that our dynamic combination scheme selects good contributions for classifiers for different cases and improves tracking accuracy significantly.

cs.CV↗

arXiv2013

Deep neural networks for single channel source separation

In this paper, a novel approach for single channel source separation (SCSS) using a deep neural network (DNN) architecture is introduced. Unlike previous studies in which DNN and other classifiers were used for classifying time-frequency bins to obtain hard masks for each source, we use the DNN to classify estimated source spectra to check for their validity during separation. In the training stage, the training data for the source signals are used to train a DNN. In the separation stage, the trained DNN is utilized to aid in estimation of each source in the mixed signal. Single channel source separation problem is formulated as an energy minimization problem where each source spectra estimate is encouraged to fit the trained DNN model and the mixed signal spectrum is encouraged to be written as a weighted sum of the estimated source spectra. The proposed approach works regardless of the energy scale differences between the source signals in the training and separation stages. Nonnegative matrix factorization (NMF) is used to initialize the DNN estimate for each source. The experimental results show that using DNN initialized by NMF for source separation improves the quality of the separated signal compared with using NMF for source separation.

cs.NE↗

arXiv2013

Compressed Sensing Signal Recovery via Forward-Backward Pursuit

Recovery of sparse signals from compressed measurements constitutes an l0 norm minimization problem, which is unpractical to solve. A number of sparse recovery approaches have appeared in the literature, including l1 minimization techniques, greedy pursuit algorithms, Bayesian methods and nonconvex optimization techniques among others. This manuscript introduces a novel two stage greedy approach, called the Forward-Backward Pursuit (FBP). FBP is an iterative approach where each iteration consists of consecutive forward and backward stages. The forward step first expands the support estimate by the forward step size, while the following backward step shrinks it by the backward step size. The forward step size is larger than the backward step size, hence the initially empty support estimate is expanded at the end of each iteration. Forward and backward steps are iterated until the residual power of the observation vector falls below a threshold. This structure of FBP does not necessitate the sparsity level to be known a priori in contrast to the Subspace Pursuit or Compressive Sampling Matching Pursuit algorithms. FBP recovery performance is demonstrated via simulations including recovery of random sparse signals with different nonzero coefficient distributions in noisy and noise-free scenarios in addition to the recovery of a sparse image.

cs.IT↗

arXiv2013

Online Recovery Guarantees and Analytical Results for OMP

Orthogonal Matching Pursuit (OMP) is a simple, yet empirically competitive algorithm for sparse recovery. Recent developments have shown that OMP guarantees exact recovery of K-sparse signals with K or more than K iterations if the observation matrix satisfies the restricted isometry property (RIP) with some conditions. We develop RIP-based online guarantees for recovery of a K-sparse signal with more than K OMP iterations. Though these guarantees cannot be generalized to all sparse signals a priori, we show that they can still hold online when the state-of-the-art K-step recovery guarantees fail. In addition, we present bounds on the number of correct and false indices in the support estimate for the derived condition to be less restrictive than the K-step guarantees. Under these bounds, this condition guarantees exact recovery of a K-sparse signal within 3K/2 iterations, which is much less than the number of steps required for the state-of-the-art exact recovery guarantees with more than K steps. Moreover, we present phase transitions of OMP in comparison to basis pursuit and subspace pursuit, which are obtained after extensive recovery simulations involving different sparse signal types. Finally, we empirically analyse the number of false indices in the support estimate, which indicates that these do not violate the developed upper bound in practice.

cs.IT↗

arXiv2013

Source Separation using Regularized NMF with MMSE Estimates under GMM Priors with Online Learning for The Uncertainties

We propose a new method to enforce priors on the solution of the nonnegative matrix factorization (NMF). The proposed algorithm can be used for denoising or single-channel source separation (SCSS) applications. The NMF solution is guided to follow the Minimum Mean Square Error (MMSE) estimates under Gaussian mixture prior models (GMM) for the source signal. In SCSS applications, the spectra of the observed mixed signal are decomposed as a weighted linear combination of trained basis vectors for each source using NMF. In this work, the NMF decomposition weight matrices are treated as a distorted image by a distortion operator, which is learned directly from the observed signals. The MMSE estimate of the weights matrix under GMM prior and log-normal distribution for the distortion is then found to improve the NMF decomposition results. The MMSE estimate is embedded within the optimization objective to form a novel regularized NMF cost function. The corresponding update rules for the new objectives are derived in this paper. Experimental results show that, the proposed regularized NMF algorithm improves the source separation performance compared with using NMF without prior or with other prior models.

cs.LG↗

arXiv2012

A* Orthogonal Matching Pursuit: Best-First Search for Compressed Sensing Signal Recovery

Compressed sensing is a developing field aiming at reconstruction of sparse signals acquired in reduced dimensions, which make the recovery process under-determined. The required solution is the one with minimum $\ell_0$ norm due to sparsity, however it is not practical to solve the $\ell_0$ minimization problem. Commonly used techniques include $\ell_1$ minimization, such as Basis Pursuit (BP) and greedy pursuit algorithms such as Orthogonal Matching Pursuit (OMP) and Subspace Pursuit (SP). This manuscript proposes a novel semi-greedy recovery approach, namely A* Orthogonal Matching Pursuit (A*OMP). A*OMP performs A* search to look for the sparsest solution on a tree whose paths grow similar to the Orthogonal Matching Pursuit (OMP) algorithm. Paths on the tree are evaluated according to a cost function, which should compensate for different path lengths. For this purpose, three different auxiliary structures are defined, including novel dynamic ones. A*OMP also incorporates pruning techniques which enable practical applications of the algorithm. Moreover, the adjustable search parameters provide means for a complexity-accuracy trade-off. We demonstrate the reconstruction ability of the proposed scheme on both synthetically generated data and images using Gaussian and Bernoulli observation matrices, where A*OMP yields less reconstruction error and higher exact recovery frequency than BP, OMP and SP. Results also indicate that novel dynamic cost functions provide improved results as compared to a conventional choice.

cs.IT↗

arXiv2011

Max-Margin Stacking and Sparse Regularization for Linear Classifier Combination and Selection

The main principle of stacked generalization (or Stacking) is using a second-level generalizer to combine the outputs of base classifiers in an ensemble. In this paper, we investigate different combination types under the stacking framework; namely weighted sum (WS), class-dependent weighted sum (CWS) and linear stacked generalization (LSG). For learning the weights, we propose using regularized empirical risk minimization with the hinge loss. In addition, we propose using group sparsity for regularization to facilitate classifier selection. We performed experiments using two different ensemble setups with differing diversities on 8 real-world datasets. Results show the power of regularized learning with the hinge loss function. Using sparse regularization, we are able to reduce the number of selected classifiers of the diverse ensemble without sacrificing accuracy. With the non-diverse ensembles, we even gain accuracy on average by using sparse regularization.

cs.LG↗