Exploiting Foundation Models and Speech Enhancement for Parkinson's Disease Detection from Speech in Real-World Operative Conditions

• DOI: 10.21437/interspeech.2024-522
• Publication Date: 2024-09-01T07:10:12Z
• Authors: Moreno La Quatra; Maria Francesca T...; Torbjørn Svendsen; Giampiero Salvi; Juan Rafael Orozc...; Sabato Marco Sini...
• Keywords: Foundation (evidence)

This work is concerned with devising a robust Parkinson's (PD) disease detector from speech in real-world operating conditions using (i) foundational models, and (ii) enhancement (SE) methods. To this end, we first fine-tune several foundational-based models on the standard PC-GITA (s-PC-GITA) clean data. Our results demonstrate superior performance to previously proposed models. Second, assess generalization capability of PD extended (e-PC-GITA) recordings, collected operative conditions, observe severe drop moving ideal conditions. Third, align training testing applaying off-the-shelf SE techniques e-PC-GITA, significant boost observed only for Finally, combining two best trained s-PC-GITA, namely WavLM Base Hubert Base, yielded top enhanced e-PC-GITA.