The effect of reverberation on the performance of cepstral mean subtraction in speaker verification

Speaker verification (SVR) performance is degraded under reverberation conditions. Cepstral mean subtraction (CMS) is often applied to the feature vectors in order to compensate for convolutive effects of transmission channels, which are considered to have a short-duration impulse response. The effect of reverberation on the performance of CMS applied to the feature vectors in SVR is investigated. Although CMS was found effective in reducing the effect of reverberation for short reverberation time (RT), in cases of long RT, it is shown that CMS may degrade SVR performance rather than improve it. Hence, CMS should not to be used in these cases. In addition, the effect of the room volume was tested and found less critical than the effect of long RT.     

The physics of categorization

In the context of cognition, categorization is the process through which several elements (i.e., words) are grouped into a single set which by naming becomes an abstraction of its elements. For example, tiger, kitty, and max can be categorized as Cats. In this article, we aim to show how the physical, biological and cognitive dimensions are related in the process of categorization or abstraction through the physics of computation. Drawing on Landauer's principle, we show that the price paid in terms of entropy is higher when grouping elements of low ranking (high probability) than when grouping elements of high ranking (low probability). Therefore, the logic of the cognitive process of abstraction is explained through constraints imposed by memory on the computation of categories. © 2015 Wiley Periodicals, Inc. Complexity 21: 269–274, 2016     

Acoustic analysis by spherical microphone array processing of room impulse responses

Spherical microphone arrays have been recently used for room acoustics analysis, to detect the direction-of-arrival of early room reflections, and compute directional room impulse responses and other spatial room acoustics parameters. Previous works presented methods for room acoustics analysis using spherical arrays that are based on beamforming, e.g., delay-and-sum, regular beamforming, and Dolph-Chebyshev beamforming. Although beamforming methods provide useful directional selectivity, optimal array processing methods can provide enhanced performance. However, these algorithms require an array cross-spectrum matrix with a full rank, while array data based on room impulse responses may not satisfy this condition due to the single frame data. This paper presents a smoothing technique for the cross-spectrum matrix in the frequency domain, designed for spherical microphone arrays, that can solve the problem of low rank when using room impulse response data, therefore facilitating the use of optimal array processing methods. Frequency smoothing is shown to be performed effectively using spherical arrays, due to the decoupling of frequency and angular components in the spherical harmonics domain. Experimental study with data measured in a real auditorium illustrates the performance of optimal array processing methods such as MUSIC and MVDR compared to beamforming.     

Null sets and games in Banach spaces

The notion of Aronszajn-null sets generalizes the notion of Lebesgue measure zero in the Euclidean space to infinite dimensional Banach spaces. We present a game-theoretic approach to Aronszajn-null sets, establish its basic properties, and discuss some ensuing open problems.     

The combinatorics of the Baer-Specker group

We study subgroups of ℤ^ℕ which possess group theoretic properties analogous to properties introduced by Menger (1924), Hurewicz (1925), Rothberger (1938), and Scheepers (1996). The studied properties were introduced independently by Kočinac and Okunev. We obtain purely combinatorial characterizations of these properties, and combine them with other techniques to solve several questions of Babinkostova, Kočinac, and Scheepers. The first author was supported by the EU Research and Training Network HPRN-CT-2002-00287. The second author was partially supported by the Koshland Center for Basic Research.     

High-precision measurements of 17O/16O and 18O/16O of O2 and O2/Ar ratio in air

A method for high-precision and high-accuracy mass spectrometric measurements of the ratios among the three oxygen isotopes, and of the O(2)/Ar ratio, is presented. It involves separation of the O(2)-Ar mixture from air and includes a fully automated system that ensures highly reliable sample processing. Repeated measurements of atmospheric oxygen yield the repeatability (+/-SE x t, standard error of the mean (n = 12) multiplied by Student's t-factor for a 95% confidence limit) of 0.004, 0.003 and 0.2 per thousand for delta(18)O, delta(17)O and delta O(2)/Ar, respectively.     

d-orbital stereoelectronic control of the stereochemistry of SN2′ displacements by organocuprate reagents

The $\text{anti}$ stereochemistry observed in organocuprate S_N2′ displacements can be rationalized as a stereoelectronic effect arising from “bidentate” binding involving a d orbital of nueleophilic copper and π* and σ* orbitals of the substrate. The extension of this idea to other reactions of organocuprates including additions to acetylenes and enones is discussed.     

Permutation graphs, fast forward permutations, and sampling the cycle structure of a permutation

PS_N is a fast forward permutation if for each m the computational complexity of evaluating P^m(x) is small independently of m and x. Naor and Reingold constructed fast forward pseudorandom cycluses and involutions. By studying the evolution of permutation graphs, we prove that the number of queries needed to distinguish a random cyclus from a random permutation in S_N is Θ(N) if one does not use queries of the form P^m(x), but is only Θ(1) if one is allowed to make such queries. We construct fast forward permutations which are indistinguishable from random permutations even when queries of the form P^m(x) are allowed. This is done by introducing an efficient method to sample the cycle structure of a random permutation, which in turn solves an open problem of Naor and Reingold.