arXiv subjects

Hesham Ali

Publications and source records attributed to Hesham Ali.

3 recordsLinked to original sources

arXiv2026

LLM-to-Speech: A Synthetic Data Pipeline for Training Dialectal Text-to-Speech Models

Despite the advances in neural text to speech (TTS), many Arabic dialectal varieties remain marginally addressed, with most resources concentrated on Modern Spoken Arabic (MSA) and Gulf dialects, leaving Egyptian Arabic -- the most widely understood Arabic dialect -- severely under-resourced. We address this gap by introducing NileTTS: 38 hours of transcribed speech from two speakers across diverse domains including medical, sales, and general conversations. We construct this dataset using a novel synthetic pipeline: large language models (LLM) generate Egyptian Arabic content, which is then converted to natural speech using audio synthesis tools, followed by automatic transcription and speaker diarization with manual quality verification. We fine-tune XTTS v2, a state-of-the-art multilingual TTS model, on our dataset and evaluate against the baseline model trained on other Arabic dialects. Our contributions include: (1) the first publicly available Egyptian Arabic TTS dataset, (2) a reproducible synthetic data generation pipeline for dialectal TTS, and (3) an open-source fine-tuned model. All resources are released to advance Egyptian Arabic speech synthesis research.

cs.CL↗

arXiv2025

A County-Level Similarity Network of Electric Vehicle Adoption: Integrating Predictive Modeling and Graph Theory

Electric vehicle (EV) adoption is essential for reducing carbon dioxide (CO2) emissions from internal combustion engine vehicles (ICEVs), which account for nearly half of transportation-related emissions in the United States. Yet regional EV adoption varies widely, and prior studies often overlook county-level heterogeneity by relying on broad state-level analyses or limited city samples. Such approaches risk masking local patterns and may lead to inaccurate or non-transferable policy recommendations. This study introduces a graph-theoretic framework that complements predictive modeling to better capture how county-level characteristics relate to EV adoption. Feature importances from multiple predictive models are averaged and used as weights within a weighted Gower similarity metric to construct a county similarity network. A mutual k-nearest-neighbors procedure and modularity-based community detection identify 27 clusters of counties with similar weighted feature profiles. EV adoption rates are then analyzed across clusters, and standardized effect sizes (Cohens d) highlight the most distinguishing features for each cluster. Findings reveal consistent global trends, such as declining median income, educational attainment, and charging-station availability across lower adoption tiers; while also uncovering important local variations that general trend or prediction analyses fail to capture. In particular, some low-adoption groups are rural but not economically disadvantaged, whereas others are urbanized yet experience high poverty rates, demonstrating that different mechanisms can lead to the same adoption outcome. By exposing both global structural patterns and localized deviations, this framework provides policymakers with actionable, cluster-specific insights for designing more effective and context-sensitive EV adoption strategies.

cs.CY↗

arXiv2023

Two-stage Joint Transductive and Inductive learning for Nuclei Segmentation

AI-assisted nuclei segmentation in histopathological images is a crucial task in the diagnosis and treatment of cancer diseases. It decreases the time required to manually screen microscopic tissue images and can resolve the conflict between pathologists during diagnosis. Deep Learning has proven useful in such a task. However, lack of labeled data is a significant barrier for deep learning-based approaches. In this study, we propose a novel approach to nuclei segmentation that leverages the available labelled and unlabelled data. The proposed method combines the strengths of both transductive and inductive learning, which have been previously attempted separately, into a single framework. Inductive learning aims at approximating the general function and generalizing to unseen test data, while transductive learning has the potential of leveraging the unlabelled test data to improve the classification. To the best of our knowledge, this is the first study to propose such a hybrid approach for medical image segmentation. Moreover, we propose a novel two-stage transductive inference scheme. We evaluate our approach on MoNuSeg benchmark to demonstrate the efficacy and potential of our method.

eess.IV↗