This repository contains two machine learning projects focused on music and face recognition. Each project is detailed below.
This project explores various aspects of digital sound processing and machine learning applied to music data. The primary goals include creating digital sound with Python, understanding signal processing through Fourier transform, and utilizing machine learning for music genre classification.
- Basics of digital sound processing
- Signal processing using Fourier transform
- Machine learning for music genre classification
- Deep learning with PyTorch
The project includes code snippets and explanations for:
- Sampling rate effects on sound
- Fast Fourier Transform (FFT) for frequency analysis
- Creating mono-frequency sound waves
- Melody generation and contamination with noise
- Denoising using FFT
- Short Term Fourier Transform (spectrogram visualization)
- Music genre classification using spectrograms
- PCA analysis and logistic regression
- SVM classification with grid search
- Introduction to deep learning with PyTorch
This project delves into the concept of similarity in unsupervised learning, specifically applied to face recognition. The goal is to understand the difference between distance and similarity and how it can be used in facial image recognition.
- Understanding distance and similarity in unsupervised learning
- Face recognition using similarity vectors
- Normalizing vectors for unit norms
- Computing Euclidean distance for similarity
- Application of similarity concepts to facial image recognition
- Text-to-vector conversion for document similarity
- Visualization of vectors and similarity concepts
- Normalization and visualization of unit-norm vectors
- Calculation of Euclidean distance for similarity scores
Feel free to explore the individual project folders for more detailed information and code implementations.