Ceres performs spectral analysis of a sound, and presents this analysis as a sonogram where the x axis represents time, and the y-axis frequency and amplitude appears as a color.

Various operations can then be performed in the frequency domain, such as pitch-shift, spectrum shift, time-stretch, filtration, removal of overtones selected from either amplitude or coincidence, the management of partials to musical scales, etc. The parameters may change over time. Analyzed data (amplitude, pitch, centroid) can control the algorithms. Ceres may export data to Csound and other programs.

It is in many ways more intuitive to work with sound represented in the frequency domain than in the time domain. Most sound editors works in a time-domain representation, and this realy gives an ufitting image of the sound’s timbre.

Ceres is based on FFT (Fast Fourier Transform), done as blockwise spectral analysis with a fixed number of analysis channels, evenly spread along the frequency axis. To estimate the exact frequencies of partials, phase information is used. Ceres is thus a so-called phase vocoder.

Ceres was designed and programmed by Øyvind Hammer, later maintained by Kjetil S. Matheussen.