Search

Physics-Informed Neural Networks (PINNs) are used to study collision process of soliton solutions in the 2+1 dimensional Zakharov-Kuznetsov equation. The Zakharov-Kuznetsov equation originally was the model in three dimensions of plasma with a uniform magnetic field and it is a direct extension of the KdV equation into higher dimensions and is a typical quasi-integrable system. PINNs successfully solve the equations in the forward process and the solutions are obtained using the mesh-free approach and automatic differentiation while accounting for conservation laws. In the inverse process, the proper form of the equation can be successfully derived from a given training data. On the other hand, the situation becomes weird in the collision process. The result in the forward analysis no longer possesses the conservation laws, which is ought to be referred to as a dynamically incompatible field configuration (DIFC) instead of the solutions of the system. Conservative PINNs are thus introduced and we succeed to obtain the solutions that satisfy the conservation laws. The inverse analysis suggests a different equation where the coefficients exhibit the significant changes, which implies an emergent temporal interaction. With these modulated coefficients, we recalculate the equation and confirm that the conservation law has unquestionably improved., Comment: 28 pages, 16 figures, 1 Table. arXiv admin note: text overlap with arXiv:2410.19014

Spatiotemporal patterns, which are of interest in statistical physics and nonlinear dynamics, form on the tape-peeling trace. Recently, we have proposed a mathematical model to describe these pattern formation in the tape-peeling trace. In this paper, we further investigate the tape-peeling model from the perspective of its universality class. We confirm that our model belongs to the 1-dimensional directed percolation universality class. Furthermore, the experimental results from a previous study are re-analyzed, and it is suggested that the tape-peeling trace can also be classified within the 1-dimensional directed percolation universality class., Comment: 11 pages, 4 figures

Accurate estimation of plant skeletal structure (e.g., branching structure) from images is essential for smart agriculture and plant science. Unlike human skeletons with fixed topology, plant skeleton estimation presents a unique challenge, i.e., estimating arbitrary tree graphs from images. While recent graph generation methods successfully infer thin structures from images, it is challenging to constrain the output graph strictly to a tree structure. To this problem, we present TreeFormer, a plant skeleton estimator via tree-constrained graph generation. Our approach combines learning-based graph generation with traditional graph algorithms to impose the constraints during the training loop. Specifically, our method projects an unconstrained graph onto a minimum spanning tree (MST) during the training loop and incorporates this prior knowledge into the gradient descent optimization by suppressing unwanted feature values. Experiments show that our method accurately estimates target plant skeletal structures for multiple domains: Synthetic tree patterns, real botanical roots, and grapevine branches. Our implementations are available at https://github.com/huntorochi/TreeFormer/., Comment: IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2025)

Neural vocoders often struggle with aliasing in latent feature spaces, caused by time-domain nonlinear operations and resampling layers. Aliasing folds high-frequency components into the low-frequency range, making aliased and original frequency components indistinguishable and introducing two practical issues. First, aliasing complicates the waveform generation process, as the subsequent layers must address these aliasing effects, increasing the computational complexity. Second, it limits extrapolation performance, particularly in handling high fundamental frequencies, which degrades the perceptual quality of generated speech waveforms. This paper demonstrates that 1) time-domain nonlinear operations inevitably introduce aliasing but provide a strong inductive bias for harmonic generation, and 2) time-frequency-domain processing can achieve aliasing-free waveform synthesis but lacks the inductive bias for effective harmonic generation. Building on this insight, we propose Wavehax, an aliasing-free neural WAVEform generator that integrates 2D convolution and a HArmonic prior for reliable Complex Spectrogram estimation. Experimental results show that Wavehax achieves speech quality comparable to existing high-fidelity neural vocoders and exhibits exceptional robustness in scenarios requiring high fundamental frequency extrapolation, where aliasing effects become typically severe. Moreover, Wavehax requires less than 5% of the multiply-accumulate operations and model parameters compared to HiFi-GAN V1, while achieving over four times faster CPU inference speed., Comment: 13 pages, 5 figures, Submitted to IEEE/ACM Trans. ASLP

Subjective speech quality assessment (SSQA) is critical for evaluating speech samples as perceived by human listeners. While model-based SSQA has enjoyed great success thanks to the development of deep neural networks (DNNs), generalization remains a key challenge, especially for unseen, out-of-domain data. To benchmark the generalization abilities of SSQA models, we present MOS-Bench, a diverse collection of datasets. In addition, we also introduce SHEET, an open-source toolkit containing complete recipes to conduct SSQA experiments. We provided benchmark results for MOS-Bench, and we also explored multi-dataset training to enhance generalization. Additionally, we proposed a new performance metric, best score difference/ratio, and used latent space visualizations to explain model behavior, offering valuable insights for future research., Comment: Submitted to Transactions on Audio, Speech and Language Processing. This work has been submitted to the IEEE for possible publication

We discuss the neutrino mass and Hubble tension solutions and examine their effects on the Redshift-Space Distortion (RSD) observations. An analysis with RSD data indicates smaller amplitude of perturbation. Including RSD data results in a slightly weaker upper limit on the neutrino mass than that derived for data without RSD, which is common in other extended models too. We have evaluated the impacts of RSD observations on some extended models, including the varying electron mass model, a time-dependent dark energy model with two parameter equations of state (EOS), and a model where the number of neutrino species is free. When we estimate the cosmological parameters for data including RSD, we found that the EOS parameter for dark energy is larger than that of the cosmological constant, and the effective number of neutrino species is smaller than the standard value, which infers a smaller present Hubble parameter $H_0$. From the viewpoint of cosmological tensions, the varying electron mass model with non-zero neutrino mass option looks promising to relax the Hubble tension and the $S_8$ tension simultaneously., Comment: 22 pages, 6 figures

Physics-Informed Neural Networks (PINNs) are used to study vortex solutions in the 2+1 dimensional nonlinear partial differential equations. These solutions include the regularized long-wave (RLW) equation and the Zakharov-Kuznetsov (ZK) equation, which are toy models of the geostrophic shallow water model in the planetary atmosphere. PINNs successfully solve these equations in the forward process and the solutions are obtained using the mesh-free approach and automatic differentiation while accounting for conservation laws. In the inverse process, the proper equations can be successfully derived from a given training data. Since these equations have a lot in common, there are situations when substantial misidentification arises during the inverse analysis. We consider PINNs with conservation laws (referred to as cPINNs), deformations of the initial profiles, and a friction approach that provides excellent discrimination of the equations to improve the identification's resolution., Comment: 31 pages, 13 figures, 10 tables; Some Typos fixed

A rational bubble is a situation in which the asset price exceeds its fundamental value defined by the present value of dividends in a rational equilibrium model. We discuss the recent development of the theory of rational bubbles attached to real assets, emphasizing the following three points. (i) There exist plausible economic models in which bubbles inevitably emerge in the sense that all equilibria are bubbly. (ii) Such models are necessarily nonstationary but their long-run behavior can be analyzed using the local stable manifold theorem. (iii) Bubbles attached to real assets can naturally and necessarily arise with economic development. Finally, we present a model with stocks and land, and show that bubbles in aggregate stock and land prices necessarily emerge.

Automatic music transcription (AMT), aiming to convert musical signals into musical notation, is one of the important tasks in music information retrieval. Recently, previous works have applied high-resolution labels, i.e., the continuous onset and offset times of piano notes, as training targets, achieving substantial improvements in transcription performance. However, there still remain some issues to be addressed, e.g., the harmonics of notes are sometimes recognized as false positive notes, and the size of AMT model tends to be larger to improve the transcription performance. To address these issues, we propose an improved high-resolution piano transcription model to well capture specific acoustic characteristics of music signals. First, we employ the Constant-Q Transform as the input representation to better adapt to musical signals. Moreover, we have designed two architectures: the first is based on a convolutional recurrent neural network (CRNN) with dilated convolution, and the second is an encoder-decoder architecture that combines CRNN with a non-autoregressive Transformer decoder. We conduct systematic experiments for our models. Compared to the high-resolution AMT system used as a baseline, our models effectively achieve 1) consistent improvement in note-level metrics, and 2) the significant smaller model size, which shed lights on future work., Comment: Accepted to APSIPA ASC 2024

Developing a robust speech emotion recognition (SER) system in noisy conditions faces challenges posed by different noise properties. Most previous studies have not considered the impact of human speech noise, thus limiting the application scope of SER. In this paper, we propose a novel two-stage framework for the problem by cascading target speaker extraction (TSE) method and SER. We first train a TSE model to extract the speech of target speaker from a mixture. Then, in the second stage, we utilize the extracted speech for SER training. Additionally, we explore a joint training of TSE and SER models in the second stage. Our developed system achieves a 14.33% improvement in unweighted accuracy (UA) compared to a baseline without using TSE method, demonstrating the effectiveness of our framework in mitigating the impact of human speech noise. Moreover, we conduct experiments considering speaker gender, showing that our framework performs particularly well in different-gender mixture., Comment: This is the preprint version of the paper accepted at APSIPA ASC 2024

In a previous paper, we introduced quasi-BPS categories for moduli stacks of semistable Higgs bundles. Under a certain condition on the rank, Euler characteristic, and weight, the quasi-BPS categories (called BPS in this case) are non-commutative analogues of Hitchin integrable systems. We proposed a conjectural equivalence between BPS categories which swaps Euler characteristics and weights. The conjecture is inspired by the Dolbeault Geometric Langlands equivalence of Donagi--Pantev, by the Hausel--Thaddeus mirror symmetry, and by the $\chi$-independence phenomenon for BPS invariants of curves on Calabi-Yau threefolds. In this paper, we show that the above conjecture holds at the level of topological K-theories. When the rank and the Euler characteristic are coprime, such an isomorphism was proved by Groechenig--Shen. Along the way, we show that the topological K-theory of BPS categories is isomorphic to the BPS cohomology of the moduli of semistable Higgs bundles., Comment: 59 pages

In anomalous sound detection, the discriminative method has demonstrated superior performance. This approach constructs a discriminative feature space through the classification of the meta-information labels for normal sounds. This feature space reflects the differences in machine sounds and effectively captures anomalous sounds. However, its performance significantly degrades when the meta-information labels are missing. In this paper, we improve the performance of a discriminative method under unlabeled conditions by two approaches. First, we enhance the feature extractor to perform better under unlabeled conditions. Our enhanced feature extractor utilizes multi-resolution spectrograms with a new training strategy. Second, we propose various pseudo-labeling methods to effectively train the feature extractor. The experimental evaluations show that the proposed feature extractor and pseudo-labeling methods significantly improve performance under unlabeled conditions., Comment: Submitted to ICASSP2025

We present the third edition of the VoiceMOS Challenge, a scientific initiative designed to advance research into automatic prediction of human speech ratings. There were three tracks. The first track was on predicting the quality of ``zoomed-in'' high-quality samples from speech synthesis systems. The second track was to predict ratings of samples from singing voice synthesis and voice conversion with a large variety of systems, listeners, and languages. The third track was semi-supervised quality prediction for noisy, clean, and enhanced speech, where a very small amount of labeled training data was provided. Among the eight teams from both academia and industry, we found that many were able to outperform the baseline systems. Successful techniques included retrieval-based methods and the use of non-self-supervised representations like spectrograms and pitch histograms. These results showed that the challenge has advanced the field of subjective speech rating prediction., Comment: Accepted to SLT2024

Developers must select a high-performance fault localization (FL) technique from available ones. A conventional approach is to try to select only one FL technique that is expected to attain high performance before debugging activity. In contrast, we propose a new approach that dynamically selects better FL techniques during debugging activity., Comment: 2 pages, 2 figures, 1 table

Ensemble learning methods have been used to enhance the reliability of defect prediction models. However, there is an inconclusive stability of a single method attaining the highest accuracy among various software projects. This work aims to improve the performance of ensemble-learning defect prediction among such projects by helping select the highest accuracy ensemble methods. We employ bandit algorithms (BA), an online optimization method, to select the highest-accuracy ensemble method. Each software module is tested sequentially, and bandit algorithms utilize the test outcomes of the modules to evaluate the performance of the ensemble learning methods. The test strategy followed might impact the testing effort and prediction accuracy when applying online optimization. Hence, we analyzed the test order's influence on BA's performance. In our experiment, we used six popular defect prediction datasets, four ensemble learning methods such as bagging, and three test strategies such as testing positive-prediction modules first (PF). Our results show that when BA is applied with PF, the prediction accuracy improved on average, and the number of found defects increased by 7% on a minimum of five out of six datasets (although with a slight increase in the testing effort by about 4% from ordinal ensemble learning). Hence, BA with PF strategy is the most effective to attain the highest prediction accuracy using ensemble methods on various projects., Comment: 6 pages, 6 figures, 6 tables

Current cosmological observations allow for deviations from the standard growth of large-scale structures in the universe. These deviations could indicate modifications to General Relativity on cosmological scales or suggest the dynamical nature of dark energy. It is important to characterize these departures in a model-independent manner to understand their significance objectively and explore their fundamental causes more generically across a wider spectrum of theories and models. In this paper, we compress the information from redshift-space distortion data into 2-3 parameters $\mu_i$, which control the ratio between the effective gravitational coupling in Poisson's equation and Newton's constant in several redshift bins in the late universe. We test the efficiency of this compression using mock final-year data from the Dark Energy Spectroscopic Instrument (DESI) and considering three different models within the class of effective field theories of dark energy. The constraints on the parameters of these models, obtained from both the direct fit to the data and the projection of the compressed parameters onto the parameters of the models, are fully consistent, demonstrating the method's good performance. Then, we apply it to current data and find hints of a suppressed matter growth in the universe at $\sim 2.7\sigma$ C.L., in full accordance with previous works in the literature. Finally, we perform a forecast with DESI data and show that the uncertainties on the parameters $\mu_1$ at $z<1$ and $\mu_2$ at $1

American income inequality, generally estimated with tax data, in the 20th century is widely recognized to have followed a U-curve, though debates persist over the extent of this curve, specifically regarding how high the peaks are and how deep the trough is. These debates focus on assumptions about defining income and handling deductions. However, the choice of interpolation methods for using tax authorities' tabular data to estimate the income of the richest centiles -- especially when no micro-files are available -- has not been discussed. This is crucial because tabular data were consistently used from 1917 to 1965. In this paper, we show that there is an alternative to the standard method of Pareto Interpolation (PI). We demonstrate that this alternative -- Maximum Entropy (ME) -- provides more accurate results and leads to significant revisions in the shape of the U-curve of income inequality.

With the advancements in singing voice generation and the growing presence of AI singers on media platforms, the inaugural Singing Voice Deepfake Detection (SVDD) Challenge aims to advance research in identifying AI-generated singing voices from authentic singers. This challenge features two tracks: a controlled setting track (CtrSVDD) and an in-the-wild scenario track (WildSVDD). The CtrSVDD track utilizes publicly available singing vocal data to generate deepfakes using state-of-the-art singing voice synthesis and conversion systems. Meanwhile, the WildSVDD track expands upon the existing SingFake dataset, which includes data sourced from popular user-generated content websites. For the CtrSVDD track, we received submissions from 47 teams, with 37 surpassing our baselines and the top team achieving a 1.65% equal error rate. For the WildSVDD track, we benchmarked the baselines. This paper reviews these results, discusses key findings, and outlines future directions for SVDD research., Comment: 6 pages, Accepted by 2024 IEEE Spoken Language Technology Workshop (SLT 2024)

We introduce quasi-BPS categories for twisted Higgs bundles, which are building blocks of the derived category of coherent sheaves on the moduli stack of semistable twisted Higgs bundles over a smooth projective curve. Under some condition (called BPS condition), the quasi-BPS categories are non-commutative analogues of Hitchin integrable systems. We begin the study of these quasi-BPS categories by focusing on a conjectural symmetry which swaps the Euler characteristic and the weight. Our conjecture is inspired by the Dolbeault Geometric Langlands equivalence of Donagi--Pantev, by the Hausel--Thaddeus mirror symmetry, and by the $\chi$-independence phenomenon for BPS invariants of curves on Calabi-Yau threefolds. We prove our conjecture in the case of rank two and genus zero. In higher genus, we prove a derived equivalence of rank two stable twisted Higgs moduli spaces as a special case of our conjecture. In a separate paper, we prove a version of our conjecture for the topological K-theory of quasi-BPS categories and we discuss the relation between quasi-BPS categories and BPS invariants of the corresponding local Calabi-Yau threefold., Comment: 64 pages

We study a particular type of Einstein warped-product manifold where the warping function must satisfy the homogeneous version of the screened Poisson equation. Under these assumptions, we show that the dimension of the manifold, the (constant negative) Ricci curvature and the screened parameter are related through a quadratic equation., Comment: 7 pages. Preprint of manuscript accepted for publication in the Contemporary Mathematics series of the American Mathematical Society - Book entitled: "Recent Advances in Differential Geometry and Related Areas" to appear in 2025

"Rational bubble", as introduced by the famous paper on money by Samuelson (1958), means speculation backed by nothing. The large subsequent rational bubble literature has identified attaching bubbles to dividend-paying assets in a natural way as an important but challenging question. Miao and Wang (2018) claim to "provide a theory of rational stock price bubbles". Contrary to their claim, the present comment proves the nonexistence of rational bubbles in the model of Miao and Wang (2018). We also clarify the precise mathematical definition and the economic meaning of "rational bubble" in an accessible way to the general audience.

We have performed time-dependent density functional theory simulations for the neutralization process of a hydrogen ion injected at 100eV onto the (110) surface of tungsten material. We have also proposed a method for evaluating the detection probability of electrons in a small region. This probability is interpreted as that of detecting hydrogen in each state: positive ion, neutral atom, and negative ion. As a result, the probabilities of detecting hydrogen after a collision as a positive ion, neutral atom, and negative ion were approximately 30 percent, 50 percent, and 20 percent, respectively., Comment: 10 pages, 6 figures, 1 table, 26-th International Conference on Plasma Surface Interaction

The neutralization process of an ion particle on a surface is key issue of the plasma-wall interaction (PWI). We investigated helium (He) ion injection onto a tungsten surface using time-dependent density functional theory (TDDFT) simulation. We developed the TDDFT code QUMASUN for this study, simulating the process of electron transfer from the surface to the He nucleus by solving the time evolution of electron wave function and the classical motion of nuclei simultaneously. Our results show that the probabilities of injected $\text{He}^{2+}$ changing into $\text{He}^{1+}$ and $\text{He}^{0}$ on the surface are approximately 40 percent and 25 percent, respectively. The captured electrons by $\text{He}^{1+}$ and $\text{He}^{0}$ predominantly occupy the 2s and 2p orbitals, corresponding to excited states. In addition, we stated some challenges for applying TDDFT to plasma-wall interactions., Comment: 4 figures, 1 table, resented in 26th International conference on plasma-surface interaction

It has recently been argued that the Hubble tension may call for a combination of both pre- and post-recombination new physics. Motivated by these considerations, we provide one of the first concrete case studies aimed at constructing such a viable combination. We consider models that have individually worked best on either end of recombination so far: a spatially uniform time-varying electron mass leading to earlier recombination (also adding non-zero spatial curvature), and a sign-switching cosmological constant inducing an AdS-to-dS transition within the $\Lambda_{\rm s}$CDM model. When confronted against Cosmic Microwave Background (CMB), Baryon Acoustic Oscillations, and Type Ia Supernovae data, we show that no combination of these ingredients can successfully solve the Hubble tension. We find that the matter density parameter $\Omega_m$ plays a critical role, driving important physical scales in opposite directions: the AdS-to-dS transition requires a larger $\Omega_m$ to maintain the CMB acoustic scale fixed, whereas the varying electron mass requires a smaller $\Omega_m$ to maintain the redshift of matter-radiation equality fixed. Despite the overall failure, we use our results to draw general model-building lessons, highlighting the importance of assessing tension-solving directions in the parameter space of new physics parameters and how these correlate with shifts in other standard parameters, while underscoring the crucial role of $\Omega_m$ in this sense., Comment: 22 pages, 7 figures, 4 tables. Fig. 7 is the summary Figure. The reason for the failure of this particular combination of models and the consequent general lessons we draw are discussed between Pages 10 and 14. v2: several additional references added, minor change to title, added Bayesian evidence computation. Version accepted for publication in Phys. Dark Univ

Fluorescent nanothermometry has revealed pronounced inhomogeneous temperature distributions within cells and has opened the field of single-cell thermal biology. However, this finding has sparked a controversial discussion known as the 10^5 gap issue, which arises from a simple heat conduction calculation suggesting such large temperature distributions should not exist inside cells. This debate has led to a hypothesis that intracellular thermal conduction is considerably slower than in water. Although various efforts have been made to measure intracellular thermal conduction, there are significant variations in the reported values. Here, we address this issue using label-free mid-infrared photothermal microscopy, which enables us to measure heat-induced temperature variations under local thermal equilibrium via refractive index changes. We measured intracellular thermal diffusivity via transient thermal decay and found that thermal diffusivity within living cells is 93-94% of water in the cytoplasm and nucleus. This result suggests that intracellular thermal conduction cannot explain the 10^5 gap issue. To investigate this issue further, we compared fluorescent nanothermometry with our label-free thermometry by measuring heat-induced temperature variations in living cells. This experiment revealed intriguing results where the fluorescent nanothermometry showed a slowly varying unexpected signal in addition to a constant temperature change. This result indicates that fluorescent nanothermometers are sensitive to additional factors on top of the temperature variation under local thermal equilibrium, raising an open question of what fluorescent nanothermometry measures.

This study investigates how surgical intervention for speech pathology (specifically, as a result of oral cancer surgery) impacts the performance of an automatic speaker verification (ASV) system. Using two recently collected Dutch datasets with parallel pre and post-surgery audio from the same speaker, NKI-OC-VC and SPOKE, we assess the extent to which speech pathology influences ASV performance, and whether objective/subjective measures of speech severity are correlated with the performance. Finally, we carry out a perceptual study to compare judgements of ASV and human listeners. Our findings reveal that pathological speech negatively affects ASV performance, and the severity of the speech is negatively correlated with the performance. There is a moderate agreement in perceptual and objective scores of speaker similarity and severity, however, we could not clearly establish in the perceptual study, whether the same phenomenon also exists in human perception., Comment: 5 pages, 2 figures, 2 tables. Accepted to Interspeech 2024

Moment retrieval aims to locate the most relevant moment in an untrimmed video based on a given natural language query. Existing solutions can be roughly categorized into moment-based and clip-based methods. The former often involves heavy computations, while the latter, due to overlooking coarse-grained information, typically underperforms compared to moment-based models. Hence, this paper proposes a novel 2-Dimensional Pointer-based Machine Reading Comprehension for Moment Retrieval Choice (2DP-2MRC) model to address the issue of imprecise localization in clip-based methods while maintaining lower computational complexity than moment-based methods. Specifically, we introduce an AV-Encoder to capture coarse-grained information at moment and video levels. Additionally, a 2D pointer encoder module is introduced to further enhance boundary detection for target moment. Extensive experiments on the HiREST dataset demonstrate that 2DP-2MRC significantly outperforms existing baseline models., Comment: Accepted by INTERSPEECH 2024

Recent singing voice synthesis and conversion advancements necessitate robust singing voice deepfake detection (SVDD) models. Current SVDD datasets face challenges due to limited controllability, diversity in deepfake methods, and licensing restrictions. Addressing these gaps, we introduce CtrSVDD, a large-scale, diverse collection of bonafide and deepfake singing vocals. These vocals are synthesized using state-of-the-art methods from publicly accessible singing voice datasets. CtrSVDD includes 47.64 hours of bonafide and 260.34 hours of deepfake singing vocals, spanning 14 deepfake methods and involving 164 singer identities. We also present a baseline system with flexible front-end features, evaluated against a structured train/dev/eval split. The experiments show the importance of feature selection and highlight a need for generalization towards deepfake methods that deviate further from training distribution. The CtrSVDD dataset and baselines are publicly accessible., Comment: Accepted by Interspeech 2024

A recent economic literature deals with models of random growth in which the size (say, log-wealth) of an individual economic agent is subject to light-tailed random additive shocks over time. The distribution of these shocks depends on a latent Markov modulator. It has been shown that if Markov modulation is irreducible then the models give rise to a distribution of sizes across agents whose upper tail resembles, in a particular sense, that of an exponential distribution. We show that while this need not be the case under reducible Markov modulation, the upper tail will nevertheless resemble that of an Erlang distribution. A novel extension of the Rothblum index theorem from an affine setting to a holomorphic one allows us to characterize the shape parameter for the relevant Erlang distribution., Comment: 24 pages, 1 figure

Baryon acoustic oscillation (BAO) is one of the important standard rulers in cosmology. The results of the latest BAO measurements by Dark Energy Spectroscopic Instrument (DESI) survey have been reported. Cosmology with the varying electron mass model and the early dark energy models are regarded as interesting models to resolve the Hubble tension. We present constraints on the varying electron mass model and early dark energy models in new DESI data as well as cosmic microwave background by Planck and the conventional BAO data from 6dF, MGS, and DR12 and supernovae light curve data into analysis. Since new DESI BAO data indicate a slightly longer sound horizon $r_dh$ than the other BAO observations, for the varying electron mass model, the larger $H_0 =69.44\pm 0.84 $ km$/$s$/$Mpc is indicated., Comment: 18 pages, 6 figures

The trend of scaling up speech generation models poses a threat of biometric information leakage of the identities of the voices in the training data, raising privacy and security concerns. In this paper, we investigate training multi-speaker text-to-speech (TTS) models using data that underwent speaker anonymization (SA), a process that tends to hide the speaker identity of the input speech while maintaining other attributes. Two signal processing-based and three deep neural network-based SA methods were used to anonymize VCTK, a multi-speaker TTS dataset, which is further used to train an end-to-end TTS model, VITS, to perform unseen speaker TTS during the testing phase. We conducted extensive objective and subjective experiments to evaluate the anonymized training data, as well as the performance of the downstream TTS model trained using those data. Importantly, we found that UTMOS, a data-driven subjective rating predictor model, and GVD, a metric that measures the gain of voice distinctiveness, are good indicators of the downstream TTS performance. We summarize insights in the hope of helping future researchers determine the goodness of the SA system for multi-speaker TTS training., Comment: 5 pages. Submitted to Signal Processing Letters. Audio sample page: https://unilight.github.io/Publication-Demos/publications/sa-tts-spl/index.html

The rapid advancement of AI-generated singing voices, which now closely mimic natural human singing and align seamlessly with musical scores, has led to heightened concerns for artists and the music industry. Unlike spoken voice, singing voice presents unique challenges due to its musical nature and the presence of strong background music, making singing voice deepfake detection (SVDD) a specialized field requiring focused attention. To promote SVDD research, we recently proposed the "SVDD Challenge," the very first research challenge focusing on SVDD for lab-controlled and in-the-wild bonafide and deepfake singing voice recordings. The challenge will be held in conjunction with the 2024 IEEE Spoken Language Technology Workshop (SLT 2024)., Comment: Evaluation plan of the SVDD Challenge @ SLT 2024