← back to projects
CONCEPT Jupyter Notebook

DSP-501 Audio AI

Environmental sound classification — comparing raw-audio vs DSP-preprocessed pipelines across SVM / RandomForest / CNN-2D on UrbanSound8K.

DSP501
PythonPyTorchlibrosascikit-learn

The hypothesis

A small CNN-2D over Mel-spectrogram features should beat a classical SVM/RandomForest stack on UrbanSound8K, but the gap is smaller than papers imply once you control for preprocessing parity. The point of the comparison is to see whether DSP preprocessing (filter banks, framing, windowing) is a load-bearing component or a habit.

Stack & architecture

What I learned

Concept-stage. Final-project bookkeeping (DSP-501) — keeping the writeup honest about what the comparison actually measures is harder than training the model.