Computational methods for underdetermined convolutive speech localization and separation via model-based sparse component analysis

Abstract

In this paper, the problem of speech source localization and separation from recordings of convolutive underdetermined mixtures is studied. The problem is cast as recovering the spatio-spectral speech information embedded in a microphone array compressed measurements of the acoustic field. A model-based sparse component analysis framework is formulated for sparse reconstruction of the speech spectra in a reverberant acoustic resulting in joint localization and separation of the individual sources. We compare and contrast the computational approaches to model-based sparse recovery exploiting spatial sparsity as well as spectral structures underlying spectrographic representation of speech signals. In this context, we explore identification of the sparsity structures at the auditory and acoustic representation spaces. The auditory structures are formulated upon the principles of structural grouping based on proximity, autoregressive correlation and harmonicity of the spectral coefficients and they are incorporated for sparse reconstruction. The acoustic structures are formulated upon the image model of multipath propagation and they are exploited to characterize the compressive measurement matrix associated with microphone array recordings.Three approaches to sparse recovery relying on combinatorial optimization, convex relaxation and Bayesian methods are studied and evaluated based on thorough experiments. The sparse Bayesian learning method is shown to yield better perceptual quality while the interference suppression is also achieved using the combinatorial approach with the advantage of offering the most efficient computational cost. Furthermore, it is demonstrated that an average autoregressive model can be learned for speech localization and exploiting the proximity structure in the form of block sparse coefficients enables accurate localization. Throughout the extensive empirical evaluation, we confirm that a large and random placement of the microphones enables significant improvement in source localization and separation performance.

Independent component analysis: An introduction

“…Figure13shows the mixed signals and the histogram for these mixture signals. As shown in the figure, the mixture signals follow all the properties that were mentioned in Section 2.1.1, where (1) source signals are more independent than mixture signals,(2) the histograms of mixture signals in Figure13are much more Gaussian than the histogram of source signals in Figure12mixtures signals (see Figure13are more complex than source signals (see Figure12)).In the projection pursuit algorithm, mixture signals are first whitened, and then the values of the first weight vector ðw 1 Þ are initialized randomly. The value of w 1 is listed in Table1.…”

“…These data can be in the form of images, stock markets, or sounds. Hence, ICA was used for extracting source signals in many applications such as medical signals [7,34], biological assays [3], and audio signals [2]. ICA is also considered as a dimensionality reduction algorithm when ICA can delete or retain a single source.…”

Exploration of Wireless Sensor Network Based on RSSI Positioning Algorithm

“…Nayak and Devulapalli (2016) proposed a cluster-based WSN localization algorithm that uses cluster structure and a global system to represent network distribution, reduces measurement error through multi-hop probability calculation, and improves node positioning accuracy [2]. Asaei et al (2017) proposed a localization method based on perceptual sparse matrix, transforming the distribution of nodes into discrete regions of sparse matrices, and using greedy algorithms and deterministic monitoring matrices to reduce the number of measurements [3]. Sicari (2015) used a global single key to encrypt nodes in the TinSec positioning system.…”

Separation of underdetermined speech mixture based on sparse Bayesian recovery

“…Then the maximum posterior technique are usually used to estimate parameters. In this section, all the signals are complex valued, which is different from the problem in [132]. We start our study from using the following complex Gaussian distribution to model g,…”

“…The sparseness of three speech signals is illustrated inFigure 6.2. In[132], it is shown that one SBL method realized by a BSBL algorithm could achieve the best perceived speech quality. However,these sparse Bayesian recovery methods are not designed for speech signals in TF domain.…”