ECE 5725 Final Project

Haoyang & Yue

Baseline Design

This is in essence an information transformation problem, that can be decomposed into the following steps:

1. Sound to Music File: microphone
2. Music File to Spectrum: transform series data into time-frequency spectrums
3. Spectrum to Pitch: detect the correspondence between spectrum and pitch
4. Pitch to Note: detect the correspondence between pitch and note
5. Note to Symbolic Graphs: for simplified notes, text printing should be fine
6. Symbolic Graphs to Printed Format: a mini-printer

Strech Goals

• Short pitch detection (impulse)
• Harmonic detection
• Real-time detection: input from the microphone and output from the Resparry pi
• Symphony
• Pop music with lyrics
• Staff notations
• Reverse procedure: input staff notations and output the music -> machine learning