Edgard-Varèse guest professorship
During my stay as Edgard-Varèse
guest professor for computer music in summer 2006
at the Electronic studio of the
Technical University of Berlin I held a lecture
entitled: Analysis, modeling and transformation of
audio signals. The topics covered are time frequency representations
of sound signals
and phase vocoder techniques for sound transformation. Some higher
level concepts as: spectral envelopes, fundamental frequency,
and sinusoidal modeling, are discussed as well.
The lecture material is provided below.
Lecture material
-
Fundamentals of discrete Fourier analysis
Introduction of the fundamental methods of spectral analysis: Fourier transform, DFT, windowing, frequency resolution. (last update 26/July/2006)
-
Analysis/resynthesis with the short time Fourier transform
The fundamental parameters of the short time Fourier transform and the possibilities to reconstruct the original signal from the modified or unmodified STFT are discussed. (last update 26/July/2006).
- Signal modifications using the STFT
Presents some of the algorithms for signal modification that are based on a STFT representation:
time invariant and time variant filtering, time stretching, transposition. (last update 26/July/2006)
- Source filter modeling and spectral envelope estimation
source filter model of speech, envelope estimation algorithms: linear prediction, discrete all pole, cepstral smoothing, true envelope, transposition with envelope preservation. (last update 25/August/2006)
- Fundamental frequency estimation
Explains some basic concepts related to fundamental frequency estimation,
concept of a periodic signal, current techniques for the estimation. (last update 26/July/2006)
- Sinusoidal modeling
motivation of the model, peak detection, sinusoidal parameter estimation
for stationary and non stationary sinusoids, partial tracking, trajectory interpolation. (last update 26/07/2006)