Incorporando Regionalismos e Sotaques em Modelos de Síntese de Fala - Corpus

This repository contains a Brazilian Portuguese speech corpus developed as part of the project “Incorporating Regionalisms and Accents into Speech Synthesis Models”. The corpus was created to support research on speech synthesis, regional linguistic variation, and accent adaptation in text-to-speech systems, with particular attention to speech varieties from Paraíba and Northeastern Brazil. The corpus was built by organizing, processing, and integrating speech data from multiple sources, including CoLingPB, the Sotaque Brasileiro dataset, and additional collaborative speech recordings collected for this project. The data was curated and standardized to support experiments involving fine-tuning of pre-existing speech synthesis checkpoints, evaluation of regional accent preservation, and comparison between base and adapted text-to-speech models. This dataset is intended for academic and research purposes, especially for studies related to Brazilian Portuguese speech synthesis, regional accents, computational linguistics, speech processing, and deep learning approaches for text-to-speech. Users of this corpus should cite the original data sources and related works used in its construction and contextualization. Suggested citations in BibTeX format: @mastersthesis{morais2026regionalismos, author = {Morais, Thomaz Diniz Pinto de}, title = {{Incorporando Regionalismos e Sotaques em Modelos de Síntese de Fala}}, year = {2026}, school = {Universidade Federal de Campina Grande}, address = {Campina Grande, PB, Brazil}, type = {Master's thesis}, note = {Programa de Pós-Graduação em Ciência da Computação}} @misc{stein2015colingpb, author = {Stein, Cirineu Cecote and others}, title = {{Corpus Linguístico da Paraíba (CoLingPB)}}, year = {2015}, institution = {Universidade Federal da Paraíba}, address = {João Pessoa, PB, Brazil}, url = {https://www.cchla.ufpb.br/colingpb/}, note = {Corpus Linguístico da Paraíba} } @dataset{milan2021sotaquebrasileiro, author = {Milan, Gabriel Gazola}, title = {{Sotaque Brasileiro}}, year = {2021}, publisher = {Zenodo}, version = {2021-09-06}, doi = {10.5281/zenodo.5466945}, url = {https://doi.org/10.5281/zenodo.5466945} } @mastersthesis{batista2019sotaques, author = {Batista, Nathalia Alves Rocha}, title = {{Estudo sobre identificação automática de sotaques regionais brasileiros baseada em modelagens estatísticas e técnicas de aprendizado de máquina}}, year = {2019}, school = {Universidade Estadual de Campinas}, address = {Campinas, SP, Brazil}, type = {Master's thesis}, url = {https://hdl.handle.net/20.500.12733/1636122} }