The Experts below are selected from a list of 318 Experts worldwide ranked by ideXlab platform
Kiyohiro Shikano - One of the best experts on this subject based on the ideXlab platform.
-
voice conversion algorithm based on gaussian mixture model with dynamic frequency warping of straight Spectrum
International Conference on Acoustics Speech and Signal Processing, 2001Co-Authors: Tomoki Toda, Hiroshi Saruwatari, Kiyohiro ShikanoAbstract:In the voice conversion algorithm based on the Gaussian Mixture Model (GMM) applied to STRAIGHT, quality of converted speech is degraded because the converted Spectrum is exceedingly smooth. We propose the GMM-based algorithm with dynamic frequency warping to avoid the over-smoothing. We also propose an addition of the weighted Residual Spectrum, which is the difference between the GMM-based converted Spectrum and the frequency-warped Spectrum, to avoid the deterioration of conversion-accuracy on speaker individuality. Results of the evaluation experiments clarify that the converted speech quality is better than that of the GMM-based algorithm, and the conversion-accuracy on speaker individuality is the same as that of the GMM-based algorithm in the proposed method with the properly-weighted Residual Spectrum.
-
ICASSP - Voice conversion algorithm based on Gaussian mixture model with dynamic frequency warping of STRAIGHT Spectrum
2001 IEEE International Conference on Acoustics Speech and Signal Processing. Proceedings (Cat. No.01CH37221), 1Co-Authors: Tomoki Toda, Hiroshi Saruwatari, Kiyohiro ShikanoAbstract:In the voice conversion algorithm based on the Gaussian Mixture Model (GMM) applied to STRAIGHT, quality of converted speech is degraded because the converted Spectrum is exceedingly smooth. We propose the GMM-based algorithm with dynamic frequency warping to avoid the over-smoothing. We also propose an addition of the weighted Residual Spectrum, which is the difference between the GMM-based converted Spectrum and the frequency-warped Spectrum, to avoid the deterioration of conversion-accuracy on speaker individuality. Results of the evaluation experiments clarify that the converted speech quality is better than that of the GMM-based algorithm, and the conversion-accuracy on speaker individuality is the same as that of the GMM-based algorithm in the proposed method with the properly-weighted Residual Spectrum.
Tomoki Toda - One of the best experts on this subject based on the ideXlab platform.
-
voice conversion algorithm based on gaussian mixture model with dynamic frequency warping of straight Spectrum
International Conference on Acoustics Speech and Signal Processing, 2001Co-Authors: Tomoki Toda, Hiroshi Saruwatari, Kiyohiro ShikanoAbstract:In the voice conversion algorithm based on the Gaussian Mixture Model (GMM) applied to STRAIGHT, quality of converted speech is degraded because the converted Spectrum is exceedingly smooth. We propose the GMM-based algorithm with dynamic frequency warping to avoid the over-smoothing. We also propose an addition of the weighted Residual Spectrum, which is the difference between the GMM-based converted Spectrum and the frequency-warped Spectrum, to avoid the deterioration of conversion-accuracy on speaker individuality. Results of the evaluation experiments clarify that the converted speech quality is better than that of the GMM-based algorithm, and the conversion-accuracy on speaker individuality is the same as that of the GMM-based algorithm in the proposed method with the properly-weighted Residual Spectrum.
-
ICASSP - Voice conversion algorithm based on Gaussian mixture model with dynamic frequency warping of STRAIGHT Spectrum
2001 IEEE International Conference on Acoustics Speech and Signal Processing. Proceedings (Cat. No.01CH37221), 1Co-Authors: Tomoki Toda, Hiroshi Saruwatari, Kiyohiro ShikanoAbstract:In the voice conversion algorithm based on the Gaussian Mixture Model (GMM) applied to STRAIGHT, quality of converted speech is degraded because the converted Spectrum is exceedingly smooth. We propose the GMM-based algorithm with dynamic frequency warping to avoid the over-smoothing. We also propose an addition of the weighted Residual Spectrum, which is the difference between the GMM-based converted Spectrum and the frequency-warped Spectrum, to avoid the deterioration of conversion-accuracy on speaker individuality. Results of the evaluation experiments clarify that the converted speech quality is better than that of the GMM-based algorithm, and the conversion-accuracy on speaker individuality is the same as that of the GMM-based algorithm in the proposed method with the properly-weighted Residual Spectrum.
Neven Grbac - One of the best experts on this subject based on the ideXlab platform.
-
On the Residual Spectrum of split classical groups supported in the Siegel maximal parabolic subgroup
Monatshefte für Mathematik, 2011Co-Authors: Neven GrbacAbstract:For the split symplectic and special orthogonal groups over a number field, we decompose the part of the Residual Spectrum supported in the maximal parabolic subgroup with the Levi factor isomorphic to GL _ n . The decomposition depends on the analytic properties of the symmetric and exterior square automorphic L-functions, but seems sufficient for the computation of the corresponding part of the Eisenstein cohomology. We also prove that if one assumed Arthur’s conjectural description of the discrete Spectrum for the considered groups, then one would be able to find the poles of the L-functions in question, and would make the decomposition more precise.
-
Residual Spectra of Split Classical Groups and their Inner Forms
Canadian Journal of Mathematics, 2009Co-Authors: Neven GrbacAbstract:This paper is concerned with the Residual Spectrum of the hermitian quaternionic classical groups and defined as algebraic groups for a quaternion algebra over an algebraic number field. Groups and are not quasi-split. They are inner forms of the split groups and . Hence, the parts of the Residual Spectrum of and obtained in this paper are compared to the corresponding parts for the split groups and .
-
On the Residual Spectrum of Hermitian quaternionic inner form of SO_8
Glasnik Matematicki, 2009Co-Authors: Neven GrbacAbstract:In this paper we decompose the Residual Spectrum sup- ported in the minimal parabolic subgroup of an inner form of the split group SO8. The approach is the Langlands spectral theory. However, since the group is non-quasi-split, it is out of the scope of the Langlands-Shahidi method and the new technique for the normalization of standard intertwin- ing operators is developed. The decomposition shows interesting parts of the Residual Spectrum not appearing in the case of quasi-split groups.
-
The Residual Spectrum of an inner form of $Sp_8$ supported in the minimal parabolic subgroup
Transactions of the American Mathematical Society, 2009Co-Authors: Neven GrbacAbstract:The part of the Residual Spectrum of an inner form of the split group Sp 8 supported in the minimal parabolic subgroup is decomposed. Since the considered inner form is not quasi-split, the normalization of the standard intertwining operators, required for the calculation of the poles of the Eisenstein series, is out of the reach of the Langlands-Shahidi method. Hence, a normalization technique, based on the transfer of the Plancherel measure between the split group and its inner form, is applied. The obtained decomposition reveals certain features of the Residual Spectrum of the inner form which do not appear for the split group.
-
Global Jacquet–Langlands correspondence, multiplicity one and classification of automorphic representations
Inventiones Mathematicae, 2008Co-Authors: Alexandru Ioan Badulescu, Neven GrbacAbstract:In this paper we generalize the local Jacquet-Langlands correspondence to all unitary irreducible representations. We prove the global Jacquet-Langlands correspondence in characteristic zero. As consequences we obtain the multiplicity one and strong multiplicity one Theorems for inner forms of GL(n) as well as a classification of the Residual Spectrum and automorphic representations in analogy with results proved by Moeglin-Waldspurger and Jacquet-Shalika for GL(n).
Hiroshi Saruwatari - One of the best experts on this subject based on the ideXlab platform.
-
voice conversion algorithm based on gaussian mixture model with dynamic frequency warping of straight Spectrum
International Conference on Acoustics Speech and Signal Processing, 2001Co-Authors: Tomoki Toda, Hiroshi Saruwatari, Kiyohiro ShikanoAbstract:In the voice conversion algorithm based on the Gaussian Mixture Model (GMM) applied to STRAIGHT, quality of converted speech is degraded because the converted Spectrum is exceedingly smooth. We propose the GMM-based algorithm with dynamic frequency warping to avoid the over-smoothing. We also propose an addition of the weighted Residual Spectrum, which is the difference between the GMM-based converted Spectrum and the frequency-warped Spectrum, to avoid the deterioration of conversion-accuracy on speaker individuality. Results of the evaluation experiments clarify that the converted speech quality is better than that of the GMM-based algorithm, and the conversion-accuracy on speaker individuality is the same as that of the GMM-based algorithm in the proposed method with the properly-weighted Residual Spectrum.
-
ICASSP - Voice conversion algorithm based on Gaussian mixture model with dynamic frequency warping of STRAIGHT Spectrum
2001 IEEE International Conference on Acoustics Speech and Signal Processing. Proceedings (Cat. No.01CH37221), 1Co-Authors: Tomoki Toda, Hiroshi Saruwatari, Kiyohiro ShikanoAbstract:In the voice conversion algorithm based on the Gaussian Mixture Model (GMM) applied to STRAIGHT, quality of converted speech is degraded because the converted Spectrum is exceedingly smooth. We propose the GMM-based algorithm with dynamic frequency warping to avoid the over-smoothing. We also propose an addition of the weighted Residual Spectrum, which is the difference between the GMM-based converted Spectrum and the frequency-warped Spectrum, to avoid the deterioration of conversion-accuracy on speaker individuality. Results of the evaluation experiments clarify that the converted speech quality is better than that of the GMM-based algorithm, and the conversion-accuracy on speaker individuality is the same as that of the GMM-based algorithm in the proposed method with the properly-weighted Residual Spectrum.
Alatancang Chen - One of the best experts on this subject based on the ideXlab platform.
-
The point Spectrum and Residual Spectrum of upper triangular operator matrices
Filomat, 2019Co-Authors: Jun-jie Huang, Alatancang ChenAbstract:The point and Residual spectra of an operator are, respectively, split into 1,2-point Spectrum and 1,2-Residual Spectrum, based on the denseness and closedness of its range. Let H,K be infinite dimensional complex separable Hilbert spaces and write MX = (AX0B) ? B(H?K). For given operators A ? B(H) and B ? B(K), the sets ? X?B(K,H) ?+,i(MX)(+ = p,r;i = 1,2), are characterized. Moreover, we obtain some necessary and sufficient condition such that ?*,i(MX) = ?*,i(A) ?*,i(B) (* = p,r;i = 1,2) for every X ? B(K,H).
-
Spectra of 2 × 2 Upper Triangular Operator Matrices
Filomat, 2016Co-Authors: Jun-jie Huang, Aichun Liu, Alatancang ChenAbstract:The spectra of the $2\times 2$ upper triangular operator matrix $M_C =\big({A\ C\atop 0\ B}\big)$ acting on a Hilbert space $H_1\oplus H_2$ are investigated. We obtain a necessary and sufficient condition of $\sg(M_C)=\sg(A)\cup\sg(B)$ for every $C\in\calB(H_2,H_1)$, in terms of the spectral properties of two diagonal elements $A$ and $B$ of $M_C$. Also, the analogues for the point Spectrum, Residual Spectrum and continuous Spectrum are further presented. Moveover, we construct some examples illustrating our main results. In particular, it is shown that the inclusion $\sg_r(M_C)\subseteq\sg_r(A)\cup \sg_r(B)$ for every $C\in\calB(H_2,H_1)$ is not correct in general. Note that $\sg(T)$ (resp.\,$\sg_r(T)$) denotes the Spectrum (resp. Residual Spectrum) of an operator $T$, and $\calB(H_2,H_1)$ is the set of all bounded linear operators from $H_2$ to $H_1$.
-
Spectra of 2 x 2 upper triangular operator matrices
Filomat, 2016Co-Authors: Jun-jie Huang, Aichun Liu, Alatancang ChenAbstract:The spectra of the 2 x 2 upper triangular operator matrix MC = (A C 0 B ) acting on a Hilbert space H1 ? H2 are investigated. We obtain a necessary and sufficient condition of ?(MC) = ?(A)??(B) for every C ? B(H2,H1), in terms of the spectral properties of two diagonal elements A and B of MC. Also, the analogues for the point Spectrum, Residual Spectrum and continuous Spectrum are further presented. Moveover, we construct some examples illustrating our main results. In particular, it is shown that the inclusion ?r(MC) ? ?r(A) ? ?r(B) for every C ? B(H2,H1) is not correct in general. Note that ?(T) (resp. ?r(T)) denotes the Spectrum (resp. Residual Spectrum) of an operator T, and B(H2,H1) is the set of all bounded linear operators from H2 to H1.
-
The Residual Spectrum and the Continuous Spectrum of Upper Triangular Operator Matrices
Filomat, 2014Co-Authors: Guojun Hai, Alatancang ChenAbstract:Let H and K be separable infinite dimensional Hilbert spaces. We denote by MC the 2 2 upper triangular operator matrix acting on H K of the form MC = ( A C c(MC) are characterized, where σr( ) and σc( ) denote the Residual Spectrum and the continuous Spectrum, respectively.
-
Spectral properties and numerical range of off-diagonal infinite-dimensional Hamiltonian operators
Linear and Multilinear Algebra, 2012Co-Authors: Alatancang ChenAbstract:The spectral properties of the off-diagonal infinite-dimensional Hamiltonian operator are studied in this article. Characterizations of the Spectrum, the point Spectrum and the Residual Spectrum are given. A spectral inclusion property of numerical range is proved.