Sound Examples

The following are some sound examples which make use of some of the theory developed in Music 421 / EE 367B. The first group of sound examples are by Mark Dolson. For more information, look at his excellent tutorial on phase vocoders (in the Reader).

Example 1 Time Expansion
Example 2 Time Expansion ( Flute )

The second group of sound examples are by Celso Aguiar . Celso made these using Xavier Serra's Spectral Modeling Synthesis (SMS) software. Basically, this technique assumes a signal model based on a sinusoidal (deterministic)+ noise (stochastic) representation. The Short Time Fourier Transform is used for analysis purposes to extract and track the major sinusoidal peaks. For these sounds, the analysis was carried out with Xavier's programs. The resynthesis uses Celso's own programs combining Xavier Rodet's IFFT algorithm for additive synthesis in CLM-C-Lisp's compositional environment. In SMS, the Stochastic representation is modeled by performing a spectral fitting on the residual signal. For these sounds, the residual itself was employed, phase-vocodered whenever necessary.

This signal representation is very flexible. It allows for time and frequency modifications to be made to the signal. Additionally it can be used for `morphing' of sounds.

original Cello sound - A3 open-string.
modified Cello 1 - Spectrum is transformed in sound's second half. Fundamental is maintained.
modified Cello 2 - Spectral-domain (sine-wise) granulation and transformation in sound's second half.
modified Cello 3 - Sound is stretched and transformed in its first half.
modified Cello 4 - Transformed first half of cello-sound.
piano to cello morphing - Spectral interpolation from piano to cello (C2 note).
cymbal to voice morphing - Spectral interpolation from cymbal to singing voice. (from "Piece of Mind" by the same).

The third group of sounds were realized by Juan Pampin. They are the results of cross synthesis of singing voice and several sounds of nature (water, wind and fire)

Vocal sound "ah" (singing @ a#2)

Filtered sources:

Variable filter BW:

Constant filter BW (vocal partials were shifted by 3/8 to 8/8 of f0)

These sounds are part of a collaboration of Juan Pampin with professor Jonathan Harvey.

See Scott Levine's thesis home page for some sound examples of "sines + noise + transients" modeling, including time compression-expansion and frequency-shifting examples in the compressed domain.

Back to the Music 421 class homepage.