A dataset of genuine and synthetic Portuguese speech covering European, Brazilian, African, and regional accents - An open resource for speech deepfake detection and trustworthy AI research

André Miguel Viegas Oliveira, Rafael Geraldo dos Santos, Helder Daniel, José Valente de Oliveira · 2026

Read A dataset of genuine and synthetic Portuguese speech covering European, Brazilian, African, and regional accents - An open resource for speech deepfake detection and trustworthy AI research by André Miguel Viegas Oliveira, Rafael Geraldo dos Santos, Helder Daniel, José Valente de Oliveira (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13571521.txt.

Description

André Oliveira and co-authors release a large open dataset of real and fake Portuguese speech, published in 2026. Portuguese has about three hundred million speakers but little support in fake-speech detection research. Their set holds over 242,000 short clips, split evenly between real and fake. Brazilian samples came from public collections, while European, African, and island varieties came from broadcasts and podcasts.

Fake clips were made with a modern text-to-speech system for the Brazilian variety and voice conversion for the others. All audio was cleaned to one format and split into training and test sets with labeled protocols. The authors target biometric security, speaker verification, and anti-spoofing research. They present the set as a foundation for trustworthy speech technology in the Portuguese-speaking world.

Catalog record

CREATORAndré Miguel Viegas Oliveira, Rafael Geraldo dos Santos, Helder Daniel, José Valente de Oliveira
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13571521

Browse the full library · About the Tower of Babel