Overview
The system analyses vocal biomarkers and speech patterns using Wav2Vec2 feature extraction and an SVM classifier to identify subtle linguistic and acoustic signals linked to cognitive decline.
Why it exists
The goal was to explore a more accessible, non-invasive path for early screening that could support healthcare professionals through simple voice recordings.
What I owned
I developed the ML pipeline, processed audio data, built feature extraction with Wav2Vec2, trained the SVM classifier, and converted the model to ONNX for cross-platform use.
Built with
- Python01
- Wav2Vec202
- SVM03
- ONNX04
- Audio ML05
