Audio WAVs → manifest generation → SpeechBrain training/prediction to classify Parkinson’s vs control recordings for the Italian dataset using reproducible recipes and results reporting.