Keywords
Machine Learning, Audio Classification

Technologies
Flask, scikit-learn, Plotly, librosa, OpenAI API

AI EcoSound Tutor is a highly visual, interactive web platform built with Flask that teaches how AI processes wildlife audio. It guides users through a 7-step ML pipeline: (1) Load Data via drag-and-drop or microphone with waveform/spectrogram previews, (2) Extract Features using MFCC with optional chroma and spectral extras, (3) Reduce Dimensions via PCA, t-SNE, or UMAP with A/B comparison, (4) Cluster segments using KMeans, GMM, or HDBSCAN, (5) Explore interactively with lasso selection, audio playback, and spectrograms, (6) Train Classifiers — Random Forest, SVM, Gradient Boosting, and more — with confusion matrices, feature importance charts, and model comparison, and (7) Export as CSV or HTML reports. Every step prioritises visual feedback through interactive Plotly charts and heatmaps. A Beginner's Guide, context-aware AI chatbot, and teaching tooltips support learning throughout. Built for students, researchers, and educators in bioacoustics.