Real-Time Vocal Pitch Analyzer

A professional-grade Digital Signal Processing (DSP) application that tracks vocal pitch in real-time against a reference track. This project demonstrates high-level Python engineering, from mathematical signal analysis to multi-threaded GUI development.

Engineering Highlights

• Autocorrelation Engine: Replaced basic FFT peak detection with a robust autocorrelation algorithm to accurately find fundamental frequencies ($f_0$) while successfully ignoring loud harmonics that typically cause "octave jumps".
• Multi-Threaded Architecture: Implemented a decoupled threading system using the threading library to handle high-frequency audio I/O via sounddevice without blocking the GUI's main execution loop.
• Hanning Window Integration: Applied mathematical windowing to audio buffers to prevent spectral leakage, ensuring cleaner frequency transitions.
• Signal Smoothing: Engineered a Moving Average filter to iron out micro-jitters in human vocal input, providing a professional, readable pitch curve.
• Dynamic Data Reporting: Built a post-session analytics suite using matplotlib.gridspec for complex layout management and time-series data inspection.

Modular Architecture

The project is organized into a clean, hybrid structure for maintainability:

• analyzer.py: The "Math Engine" containing autocorrelation, scoring logic, and audio loading.
• gui.py: The "Live Dashboard" handling the real-time threading and scrolling visualization.
• report_view.py: The "Analytics View" which generates the final comparative graphs and time-series table.

Features

• Live Pitch Monitoring: Real-time visual feedback comparing user input (Red) to reference audio (Green).
• Custom File Support: Integrated file dialog for loading any .wav reference track.
• Accuracy Scoring: A mathematical percentage score based on frequency variance over the duration of the session.
• Time-Series Inspection: A detailed breakdown of performance sampled at 0.5s intervals for deep data analysis.