First tests have been done with a small sequence (8 notes and a rest). Performances with errors have been computer generated, and tests have been done by grouping error types.
