Current lines of work and structuring projects connecting prosody, language processing, experimental methods, speech technologies and researcher training.
C4AI/USP project, in collaboration with IBM/FAPESP/USP, devoted to annotating and expanding Brazilian Portuguese speech resources and to developing methods for speech recognition and synthesis. My participation focuses on prosody, spontaneous speech, synthesis evaluation and student training.
CNPq project devoted to developing, evaluating and disseminating TTS technologies capable of representing the prosodic and accent diversity of Portuguese with greater naturalness and intelligibility. It combines harmonization of spontaneous-speech corpora, prosodic segmentation, synthesis models, EyeTrackingMOS and preparation for the Blizzard Challenge 2027.
CNPq Universal 2026 proposal investigating whether EEG signals recorded during listening without explicit ratings predict naturalness estimates from an independent panel and add information to contemporary audio-based models. The design compares audio, EEG and audio+EEG models and provides for open release of corpus data, protocol, code and metadata within ethical constraints.
Project associated with a CNPq Research Productivity Fellowship, level 1D, for 2024–2028. It investigates the role of acoustic and visual prosody in discourse-structure processing through behavioral tests and online methods. The research cycles on prosodic-boundary processing in discourse are experimental developments within this agenda.
PIBIC 2026–2027 project devoted to acoustic cues to prosodic boundaries, event perception and preparation for neurocognitive and multimodal studies. It combines acoustic analysis, prosodic manipulation, behavioral experiments and eye tracking.
Projeto estruturante de digitalização, anotação, arquivamento e disseminação do material do Projeto NURC. O trabalho iniciado em 2012 consolidou uma frente duradoura de speech corpora, documentação e formação de estudantes e também alimenta colaborações posteriores em processamento e tecnologias da fala.