Extracting Information From Music Signals

Por: Kadenze . en: ,


The course introduces audio signal processing concepts motivated by examples from MIR research. More specifically students will learn about spectral analysis and time-frequency representations in general, monophonic pitch estimation, audio feature extraction, beat tracking, and tempo estimation.


Session 1: Overview And Introduction To DSP 
In this session, we will cover Phasors, Sinusoids, and Complex Numbers. Session 2: Time-Frequency Representations 
In This session, we will learn about Sampling, Quantization, RMS, and Loudness. We will also cover DFT, Hilbert Spaces, and Spectrograms. Session 3: Monophonic Pitch Analysis/Autocorrelation 
Pitch vs Fundamental Frequency, Time-domain, Frequency-domain, Perceptual Models, Overview of applications (Query-by-Humming, Auto-tunining) will be covered in this session. Session 4: Audio Feature Extraction 
We will go over Spectral Features, Mel-Frequency Cepstral Coefficients, temporal aggregation, chroma and pitch profiles. Session 5: Rhythm Analysis 
This session is about Tempo estimation, beat tracking, drum transcription, pattern detection.