arXiv subjects

Pedro Benevenuto Valadares

Publications and source records attributed to Pedro Benevenuto Valadares.

5 recordsLinked to original sources

arXiv2026

Synthetic speech detection in Brazilian Portuguese through accent-related features

Leading commercial and open-source Text-to-Speech (TTS) models fail to emulate the regional phonetic diversity of Brazilian Portuguese (pt-BR). By aggregating disparate dialects into a single training distribution, they generate a synthetic "diluted" accent: a phonetic profile attempting to represent all regional distributions simultaneously, but ultimately carrying phonological ambiguity dissociated from natural socio-phonetic realizations. This work introduces a speech deepfake detection methodology combining multilingual phone recognizers with classical signal processing to extract phoneme-level features in consonantal and vocalic realizations with high geographic variance. The analysis reveals that the distributional gap over these features suffices to distinguish natural and synthetic voices through unsupervised Kernel Density Estimation, establishing dialectal inconsistency as a useful and interpretable feature for spoofing detection in pt-BR. Evaluation on pt-BR anti-spoofing datasets shows that these explainable, lightweight, low-dimensional features can boost the performance of foundation models on the task, and show generalization capabilities in a cross-dataset leave-one-out setup.

eess.AS↗

arXiv2026

Extracting accent features in spoken Brazilian Portuguese without sociolinguistic labels

Regional accent classification in Brazilian Portuguese (pt-BR) suffers from the need for reliable labeling. While large self-supervised learning (SSL) speech models are powerful, their training pipelines dilute sociophonetic information, since accent labels are generally not reliable or are not used in training objectives. This work introduces a novel workflow for feature extraction using only acoustic labels. By isolating explicit regional accent landmarks and using a phoneme-based forced aligner (ZIPA), our targeted feature set captures dialectal variance more effectively than utterance embeddings, demonstrating that localized features can outperform general-purpose architectures on accent-related tasks using minimal and objective data labels.

eess.AS↗

arXiv2025

Complex-Valued Neural Networks for Ultra-Reliable Massive MIMO

In the evolving landscape of 5G and 6G networks, the demands extend beyond high data rates, ultra-low latency, and extensive coverage, increasingly emphasizing the need for reliability. This paper proposes an ultra-reliable multiple-input multiple-output (MIMO) scheme utilizing quasi-orthogonal space-time block coding (QOSTBC) combined with singular value decomposition (SVD) for channel state information (CSI) correction, significantly improving performance over QOSTBC and traditional orthogonal STBC (OSTBC) when analyzing spectral efficiency. Although QOSTBC enhances spectral efficiency, it also increases computational complexity at the maximum likelihood (ML) decoder. To address this, a neural network-based decoding scheme using phase-transmittance radial basis function (PT-RBF) architecture is also introduced to manage QOSTBC's complexity. Simulation results demonstrate improved system robustness and performance, making this approach a potential candidate for ultra-reliable communication in next-generation networks.

eess.SP↗

arXiv2022

PCA-based Channel Estimation for MIMO Communications

In multiple-input multiple-output communications, channel estimation is paramount to keep base stations and users on track. This paper proposes a novel PCA-based-principal component analysis-channel estimation approach for MIMO orthogonal frequency division multiplexing systems. The channel frequency response is firstly estimated with the least squares method, and then PCA is used to filter only the higher singular components of the channel impulse response, which is then converted back to the frequency domain. The proposed approach is compared with the MMSE, the minimum mean square error estimation, in terms of bit error rate versus Eb/N0.

cs.IT↗

arXiv2022

Song Emotion Recognition: a Performance Comparison Between Audio Features and Artificial Neural Networks

When songs are composed or performed, there is often an intent by the singer/songwriter of expressing feelings or emotions through it. For humans, matching the emotiveness in a musical composition or performance with the subjective perception of an audience can be quite challenging. Fortunately, the machine learning approach for this problem is simpler. Usually, it takes a data-set, from which audio features are extracted to present this information to a data-driven model, that will, in turn, train to predict what is the probability that a given song matches a target emotion. In this paper, we studied the most common features and models used in recent publications to tackle this problem, revealing which ones are best suited for recognizing emotion in a cappella songs.

cs.SD↗