I wanted to run some Performance Modeling experiments by the GPO community and get your opinions and feedback. We have been doing experiments with a company for a GPO Studio feature that will allow you to enhance MIDI performance by modeling.
Here is the original flute track that was entered with notation. Before the mockup experts rush in and charge the flute sounds "synthy", bear in mind it was deliberate. The flute part was 'moused' in with notation and therefore is very quantized and has equal velocity.
In the following examples, flute tracks were processed by Performance Models using different algorithms to achieve different performance manners. The parameters affected by the model are: Pitch Wheel, Velocity, Note duration, Volume (Breath). An Auto-Phrasing Algorithm was also used. The Cello tracks were also processed with Performance Models.
The Performance Model analyzes initial MIDI data and modifies it according to algorithms performance characters and nuances. This may be particularly useful for notation users who don't have keyboards to add all necessary performance nuances.
What do you think of these examples? Should we continue to refine the process? Or forget about it? Do you think such a feature in a future version of GPO Studio will be useful?