Quick Overview

 

Tasks


 

  • A – Idiosyncratic voice signal analysis and modeling
    • A.1 – Dysphonic voice database and feature characterization
    • A.2 – Accurate harmonic analysis and modeling of natural voiced sounds
    • A.3 – Accurate vocal tract filter modeling
    • A.4 – Accurate glottal source estimation and modeling
    • A.5 – Adaptive phonetic segmentation techniques in dysphonic voice

  • B – Perceptually natural synthesis of periodic voicing components
    • B.1 – Accurate vocal tract filter reconstruction
    • B.2 – Perceptually natural voicing implantation and prosodic control

  • C – DyNaVoiceR system integration and real-time implementation

  • D – DyNaVoiceR usability tests and fine-tuning

  • E – Management

 

Milestones


 

    • M1 – A database of dysphonic and normal Portuguese phonemes and representative features (A.1), acoustic data near the vocal folds and glottal excitation models (A.4)
    • M2 – Accurate vocal tract filter models and dysphonic-normal mapping rules (A.3)
    • M3 – Harmonic parametric models and idiosyncratic features (A.2), adaptive segmentation algorithms (A.5)
    • M4 – Accurate vocal tract filter reconstruction algorithms (B.1)
    • M5 – Accurate artificial voicing implantation algorithms (B.2)
    • M6 – DyNaVoiceR proof of concept, demo and app (C and D)