EAnalysis proposes three types of similarity matrix:

- A similarity matrix computed from the image of the sonogram.
- A simple similarity matrix computed from one or two sets of data.
- A self-similarity matrix computed from one or more sets of data.

The matrix is not saved with the project, then EAnalysis needs to compute it when you launch the project. This step takes a while. Time computation depends to the duration of project and the unit duration parameter of matrix.

A similarity matrix represents the distance between two set of data. Both sets can be the same like in figure below (self-similarity). The distance results are mapped to color values to create the representation:

Step to compute the similarity matrix:

- [option] If we use several different matrix (e.g. in self-similarity matrix), EAnalysis compute distance with euclidian distance formula.
- Distance computation from two sets of data or the same set.
- Mapping from result distance to colors. The result is a greyscale matrix.
- Mapping from greyscale to colored matrix. This step is realized in real time when you select a color preset.

To have a matrix from sonogram, EAnalysis reduces the definition of sonogram image (unit duration and frequency bands parameters) and maps the color of each pixel to its numerical value (hue color). Then, EAnalysis compute the matrix like above.