Search

When it comes to authentication in speaker verification systems, not all utterances are created equal. It is essential to estimate the quality of test utterances in order to account for varying acoustic conditions. In addition to the net-speech duration of an utterance, it is observed in this paper that phonetic richness is also a key indicator of utterance quality, playing a significant role in accurate speaker verification. Several phonetic histogram based formulations of phonetic richness are explored using transcripts obtained from an automatic speaker recognition system. The proposed phonetic richness measure is found to be positively correlated with voice authentication scores across evaluation benchmarks. Additionally, the proposed measure in combination with net speech helps in calibrating the speaker verification scores, obtaining a relative EER improvement of 5.8% on the Voxceleb1 evaluation protocol. The proposed phonetic richness based calibration provides higher benefit for short utterances with repeated words., Comment: Accepted by EUSIPCO 2024

Recent progress in generative AI technology has made audio deepfakes remarkably more realistic. While current research on anti-spoofing systems primarily focuses on assessing whether a given audio sample is fake or genuine, there has been limited attention on discerning the specific techniques to create the audio deepfakes. Algorithms commonly used in audio deepfake generation, like text-to-speech (TTS) and voice conversion (VC), undergo distinct stages including input processing, acoustic modeling, and waveform generation. In this work, we introduce a system designed to classify various spoofing attributes, capturing the distinctive features of individual modules throughout the entire generation pipeline. We evaluate our system on two datasets: the ASVspoof 2019 Logical Access and the Multi-Language Audio Anti-Spoofing Dataset (MLAAD). Results from both experiments demonstrate the robustness of the system to identify the different spoofing attributes of deepfake generation systems., Comment: Accepted by INTERSPEECH 2024

We consider the uplift of co-dimension two defect solutions of seven dimensional gauged supergravity to eleven dimensions, previously found by two of the authors. The uplifted solutions are expressed as Lin-Lunin-Maldacena solutions and an infinite family of regular solutions describing holographic defects is found using the electrostatic formulation of LLM solutions., Comment: 29 pages, pdfLaTeX, 4 figures, v2 version accepted for publication in JHEP

In this note we present a solution of $N=4,d=7$ gauged supergravity which is holographically dual to a co-dimension two defect living in a six dimensional SCFT. The solution is obtained by double analytic continuation of a two charge supersymmetric black hole solution. The condition that no conical deficits are present in the bulk and on the boundary is satisfied by a one parameter family of solutions for which some holographic observables are computed., Comment: 17 pages, PDFLaTeX, 2 figures, v2 reference added

In this paper a Penrose limit is constructed for type IIB $AdS_6\times S^2\times \Sigma$ supergravity solutions. These solutions are dual to five dimensional SCFTs related to (p,q) five brane webs, which can often be described in terms of long quiver gauge theories. The null geodesic from which the Penrose limit is constructed is localized at a unique point on the two dimensional Riemann surface $\Sigma$, where the $AdS_6$ and $S^2$ metric factors are extremal. The resulting pp-wave spacetime takes a universal form. The world sheet action of the Green-Schwarz string is quadratic in the light cone gauge and the spectrum of string excitations is obtained., Comment: 25 pages, PDFLaTeX, 2 figures. v2: minor clarifications, v3: typos corrected