I-vector based speaker recognition using advanced channel compensation techniques

This paper investigates advanced channel compensation techniques for the purpose of improving i-vector speaker verification performance in the presence of high intersession variability using the NIST 2008 and 2010 SRE corpora. The performance of four channel compensation techniques: (a) weighted maximum margin criterion (WMMC), (b) source-normalized WMMC (SN-WMMC), (c) weighted linear discriminant analysis (WLDA), and; (d) source-normalized WLDA (SN-WLDA) have been investigated. We show that, by extracting the discriminatory information between pairs of speakers as well as capturing the source variation information in the development i-vector space, the SN-WLDA based cosine similarity scoring (CSS) i-vector system is shown to provide over 20% improvement in EER for NIST 2008 interview and microphone verification and over 10% improvement in EER for NIST 2008 telephone verification, when compared to SN-LDA based CSS i-vector system. Further, score-level fusion techniques are analyzed to combine the best channel compensation approaches, to provide over 8% improvement in DCF over the best single approach, (SN-WLDA), for NIST 2008 interview/ telephone enrolment-verification condition. Finally, we demonstrate that the improvements found in the context of CSS also generalize to state-of-the-art GPLDA with up to 14% relative improvement in EER for NIST SRE 2010 interview and microphone verification and over 7% relative improvement in EER for NIST SRE 2010 telephone verification.

Improving short utterance i-vector speaker verification using utterance variance modelling and compensation techniques

This paper proposes techniques to improve the performance of i-vector based speaker verification systems when only short utterances are available. Short-length utterance i-vectors vary with speaker, session variations, and the phonetic content of the utterance. Well established methods such as linear discriminant analysis (LDA), source-normalized LDA (SN-LDA) and within-class covariance normalisation (WCCN) exist for compensating the session variation but we have identified the variability introduced by phonetic content due to utterance variation as an additional source of degradation when short-duration utterances are used. To compensate for utterance variations in short i-vector speaker verification systems using cosine similarity scoring (CSS), we have introduced a short utterance variance normalization (SUVN) technique and a short utterance variance (SUV) modelling approach at the i-vector feature level. A combination of SUVN with LDA and SN-LDA is proposed to compensate the session and utterance variations and is shown to provide improvement in performance over the traditional approach of using LDA and/or SN-LDA followed by WCCN. An alternative approach is also introduced using probabilistic linear discriminant analysis (PLDA) approach to directly model the SUV. The combination of SUVN, LDA and SN-LDA followed by SUV PLDA modelling provides an improvement over the baseline PLDA approach. We also show that for this combination of techniques, the utterance variation information needs to be artificially added to full-length i-vectors for PLDA modelling.

Short utterance PLDA speaker verification using SN-WLDA and variance modelling techniques

This paper proposes a combination of source-normalized weighted linear discriminant analysis (SN-WLDA) and short utterance variance (SUV) PLDA modelling to improve the short utterance PLDA speaker verification. As short-length utterance i-vectors vary with the speaker, session variations and phonetic content of the utterance (utterance variation), a combined approach of SN-WLDA projection and SUV PLDA modelling is used to compensate the session and utterance variations. Experimental studies have found that a combination of SN-WLDA and SUV PLDA modelling approach shows an improvement over baseline system (WCCN[LDA]-projected Gaussian PLDA (GPLDA)) as this approach effectively compensates the session and utterance variations.

Cálculos AB initio de propriedades estruturais, eletrônicas e ópticas de colmpostos CaXo3 (X=Si, Ge, Sn)

In this work we present a study of structural, electronic and optical properties, at ambient conditions, of CaSiO3, CaGeO3 and CaSnO3 crystals, all of them a member of Ca-perovskite class. To each one, we have performed density functional theory ab initio calculations within LDA and GGA approximations of the structural parameters, geometry optimization, unit cell volume, density, angles and interatomic length, band structure, carriers effective masses, total and partial density of states, dielectric function, refractive index, optical absorption, reflectivity, optical conductivity and loss function. A result comparative procedure was done between LDA and GGA calculations, a exception to CaSiO3 where only LDA calculation was performed, due high computational cost that its low symmetry crystalline structure imposed. The Ca-perovskite bibliography have shown the absence of electronic structure calculations about this materials, justifying the present work

Influence of UGT1A9 intronic I399C4T polymorphism on SN-38 glucuronidation in Asian cancer patients

Genetic polymorphisms in hepatically expressed UGT1A1 and UGT1A9 contribute to the interindividual variability i-n irinotecan disposition and toxicity. We screened UGT1A1 (UGT1A1*60, g.−3140G>A, UGT1A1*28 and UGT1A1*6) and UGT1A9 (g.−118(T)9>10 and I399C>T) genes for polymorphic variants in the promoter and coding regions, and the genotypic effect of UGT1A9 I399C>T polymorphism on irinotecan disposition in Asian cancer patients was investigated. Blood samples were collected from 45 patients after administration of irinotecan as a 90 min intravenous infusion of 375 mg/m2 once in every 3 weeks. Genotypic–phenotypic correlates showed that cancer patients heterozygous or homozygous for the I399C>T allele had approximately 2-fold lower systemic exposure to SN-38 (P<0.05) and a trend towards a higher relative extent of glucuronidation (REG) of SN-38 (P>0.05). UGT1A1–1A9 diplotype analysis showed that patients harbouring the H1/H2 (TG6GT10T/GG6GT9C) diplotype had 2.4-fold lower systemic exposure to SN-38 glucuronide (SN-38G) compared with patients harbouring the H1/H5 (TG6GT10T/GG6GT10C) diplotype (P=0.025). In conclusion, this in vivo study supports the in vitro findings of Girard et al. and suggests that the UGT1A9 I399C>T variant may be an important glucuronidating allele affecting the pharmacokinetics of SN-38 and SN-38G in Asian cancer patients receiving irinotecan chemotherapy.

Weighted LDA techniques for I-vector based speaker verification

This paper introduces the Weighted Linear Discriminant Analysis (WLDA) technique, based upon the weighted pairwise Fisher criterion, for the purposes of improving i-vector speaker verification in the presence of high intersession variability. By taking advantage of the speaker discriminative information that is available in the distances between pairs of speakers clustered in the development i-vector space, the WLDA technique is shown to provide an improvement in speaker verification performance over traditional Linear Discriminant Analysis (LDA) approaches. A similar approach is also taken to extend the recently developed Source Normalised LDA (SNLDA) into Weighted SNLDA (WSNLDA) which, similarly, shows an improvement in speaker verification performance in both matched and mismatched enrolment/verification conditions. Based upon the results presented within this paper using the NIST 2008 Speaker Recognition Evaluation dataset, we believe that both WLDA and WSNLDA are viable as replacement techniques to improve the performance of LDA and SNLDA-based i-vector speaker verification.

PLDA based speaker verification with weighted LDA techniques

This paper investigates the use of the dimensionality-reduction techniques weighted linear discriminant analysis (WLDA), and weighted median fisher discriminant analysis (WMFD), before probabilistic linear discriminant analysis (PLDA) modeling for the purpose of improving speaker verification performance in the presence of high inter-session variability. Recently it was shown that WLDA techniques can provide improvement over traditional linear discriminant analysis (LDA) for channel compensation in i-vector based speaker verification systems. We show in this paper that the speaker discriminative information that is available in the distance between pair of speakers clustered in the development i-vector space can also be exploited in heavy-tailed PLDA modeling by using the weighted discriminant approaches prior to PLDA modeling. Based upon the results presented within this paper using the NIST 2008 Speaker Recognition Evaluation dataset, we believe that WLDA and WMFD projections before PLDA modeling can provide an improved approach when compared to uncompensated PLDA modeling for i-vector based speaker verification systems.

Speaker verification using I-vector features

This PhD research has provided novel solutions to three major challenges which have prevented the wide spread deployment of speaker recognition technology: (1) combating enrolment/ verification mismatch, (2) reducing the large amount of development and training data that is required and (3) reducing the duration of speech required to verify a speaker. A range of applications of speaker recognition technology from forensics in criminal investigations to secure access in banking will benefit from the research outcomes.

Interdiffusion and Growth of the Superconductor Nb3Sn in Nb/Cu(Sn) Diffusion Couples

The confusion over the growth rate of the Nb3Sn superconductor compound following the bronze technique is addressed. Furthermore, a possible explanation for the corrugated structure of the product phase in the multifilamentary structure is discussed. Kirkendall marker experiments are conducted to study the relative mobilities of the species, which also explains the reason for finding pores in the product phase layer. The movement of the markers after interdiffusion reflects that Sn is the faster diffusing species. Furthermore, different concentrations of Sn in the bronze alloy are considered to study the effect of Sn content on the growth rate. Based on the parabolic growth constant at different temperatures, the activation energy for the growth is determined.

The effect of ball milling on the melting behavior of Sn-Cu-Ag eutectic alloy

Sn-Ag-Cu (SAC) solder alloys are the best Pb free alternative for electronic industry. Since their introduction, efforts are made to improve their efficacies by tuning the processing and composition to achieve lower melting point and better wettability. Nanostructured alloys with large boundary content are known to depress the melting points of metals and alloys. In this article we explore this possibility by processing prealloyed SAC alloys close to SAC305 composition (Sn-3wt%Ag-0.5wt%Cu) by mechanical milling which results in the formation of nanostructured alloys. Pulverisette ball mill (P7) and Vibratory ball mills are used to carry out the milling of the powders at room temperature and at lower temperatures (-104 A degrees C), respectively. We report a relatively smaller depression of melting point ranging up to 5 A degrees C with respect to original alloys. The minimum grain sizes achieved and the depression of melting point are similar for both room temperature and low-temperature processed samples. An attempt has been made to rationalize the observations in terms of the basic processes occurring during the milling.

The Broad-Line Type Ic Supernoava SN 2007ru: Adding To The Diversity Of Type Ic Supernovae

Photometric and spectral evolution of the Type Ic supernova SN 2007ru until around 210 days after maximum are presented. The spectra show broad spectral features due to very high expansion velocity, normally seen in hypernovae. The photospheric velocity is higher than other normal Type Ic supernovae (SNe Ic). It is lower than SN 1998bw at similar to 8 days after the explosion, but is comparable at later epochs. The light curve (LC) evolution of SN 2007ru indicates a fast rise time of 8 +/- 3 days to B-band maximum and postmaximum decline more rapid than other broad-line SNe Ic. With an absolute V magnitude of -19.06, SN 2007ru is comparable in brightness with SN 1998bw and lies at the brighter end of the observed SNe Ic. The ejected mass of Ni-56 is estimated to be similar to 0.4 M-circle dot. The fast rise and decline of the LC and the high expansion velocity suggest that SN 2007ru is an explosion with a high kinetic energy/ejecta mass ratio (E-K/M-ej). This adds to the diversity of SNe Ic. Although the early phase spectra are most similar to those of broad-line SN 2003jd, the [O I] line profile in the nebular spectrum of SN 2007ru shows the singly peaked profile, in contrast to the doubly peaked profile in SN 2003jd. The singly peaked profile, together with the high luminosity and the high expansion velocity, may suggest that SN 2007ru could be an aspherical explosion viewed from the polar direction. Estimated oxygen abundance 12 + log(O/H) of similar to 8.8 indicates that SN 2007ru occurred in a region with nearly solar metallicity.

Cr3+ electron paramagnetic resonance study of Sn(1-x)CrxO(2) (0.00 <= x <= 0.10)

This paper reports on the liquid-helium-temperature (5 K) electron paramagnetic resonance (EPR) spectra of Cr3+ ions in the nanoparticles of SnO2 synthesized at 600 degrees C with concentrations of 0%, 0.1%, 0.5%, 1%, 1.5%, 2.0%, 2.5%, 3.0%, 5.0%, and 10%. Each spectrum may be simulated as overlap of spectra due to four magnetically inequivalent Cr3+ centers characterized by different values of the spin-Hamiltonian parameters. Three of these centers belong to Cr3+ ions in orthorhombic sites, situated near oxygen vacancies, characterized by very large zero-field splitting parameters D and E, presumably due to the presence of nanoparticles in the samples. The fourth EPR spectrum belongs to the Cr3+ ions situated at sites with tetragonal symmetry, substituting for the Sn4+ ion, characterized by a very small value of D. In addition, there appears a ferromagnetic resonance line due to oxygen defects for samples with Cr3+ concentrations of <= 2.5%. Further, in samples with Cr3+ concentrations of >2.5%, there appears an intense and wide EPR line due to the interactions among the Cr3+ ions in the clusters formed due to rather excessive doping; the intensity and width of this line increase with increasing concentration. The Cr3+ EPR spectra observed in these nanopowders very different from those in bulk SnO2 crystals.

Hydrothermal synthesis of Ba(Ti, Sn)O3 fine powders and dielectric properties of the corresponding ceramics

Fine powders of submicron-sized crystallites of BaTiO3 were prepared at 85–130°C by the hydrothermal method, starting from TiO2.ξH2O gel and Ba(OH)2 solution. The products obtained below 110°C incorporated considerable amounts of H2O and OH− in the lattice. As-prepared BaTiO3 is cubic and converts to the tetragonal phase after heat treatment at 1200°C, accompanied by the loss of residual OH− ions. Hydrothermal reaction of SnO2.ξH2O gel with Ba(OH)2 at 150–260°C gives rise to the hydrated phase, BaSn(OH)6.3H2O, due to the amphoteric nature of SnO2.ξH2O which stabilises Sn(OH)62− anions in basic media. On heating in air or releasing the pressure in situ at 260°C, BaSn(OH)6.3H2O converts to BaSnO3 through an intermediate, BaSnO(OH)4. Solid solutions of Ba(Ti,Sn)O3 are directly formed from (TiO2 + SnO2)..ξH2O gel up to 35 mol% SnO2. At higher Sn contents, the hydrothermal products are mixtures of BaSn(OH)6.3H2O and BaTiO3, which on annealing at 1000°C result in monophasic Ba(Ti,Sn)O3. The sintering characteristics and the dielectric properties of the ceramics prepared out of these fine powders are presented. The dielectric properties of fine-grained Ba(Ti,Sn)O3 ceramics are explained on the basis of the prevailing diffuse phase transition behaviour.

Surface segregation and oxidation studies of Cu-Sn and Cu-Pd alloys by x-ray photoelectron and auger spectroscopy

X-ray photoelectron and Auger spectroscopic techniques have been employed to study surface segregation and oxidation of Cu-1 at%Sn, Cu-9at%Pd and Cu-25at%Pd alloys. Both Cu-Pd(9%) and Cu-Pd(25%) alloys show segregation of Cu when heated above 500 K. The Pd concentration was reduced by 50% at 750 K compared to the bulk composition; the enthalpy of segregation of Cu is around - 6kJ/mol. Sn segregation is seen from 470 to 650 K in the Cu-Sn(1%) alloy, and a saturation plateau of Sn concentration above 650 K is observed. Surface oxidation of Cu-Sn(1%) and Cu-Pd(9%) alloys at 500 K showed the formation of Cu2O on the surface with total suppression of Sn or Pd on the respective alloy surfaces. On vacuum annealing the oxidised Cu-Sn alloy surface at 550 K, a displacement reaction 2Cu2O+Sn→4Cu+SnO2 was observed. However, under similar annealing of the oxidised Cu-Pd(9%) alloy surface at 500 K, oxide oxygen was totally desorbed leaving the Cu-Pd alloy surface clean. In the case of the Cu-Pd(25%) alloy, only dissociatively chemisorbed oxygen was seen at 500 K which desorbed at the same temperature. Oxygen spill-over from copper to palladium is suggested as the mechanism of oxygen desorption from the oxidised Cu-Pd alloy surfaces.

Comments on "Room temperature interfacial reactions in electrodeposited Au/Sn couples"

Recently, Tang et al. (Acta Mater. 56 (2008) 5818) published a paper explaining room temperature growth of the phases in the Au/Sn system. In their analysis, they considered Au as the only mobile species for all the product phases, and Sn as virtually immobile. It is shown here that this analysis is not correct, since Sn has a higher diffusion rate through the AuSn4 phase. On the other hand, the mobilities of species in the AuSn2 phase are as yet unknown.