Guitar tablature composition tools

Bend suggestion

A machine learning model is trained to identify, within a guitar tablature, the notes that are the most likely to be played as bends, depending on their context (duration, position in the bar, surrounding notes/positions, etc.). Within a guitar pattern, it can therefore suggest to the composer alternative ways of playing this pattern. The model can also be used to enrich a tablature resulting from a transcription for guitar of a musical content which was originally designed for an other instrument, without any bend possibilities, such as the piano, as shown in the video example below.

Rhythm guitar chord transformation

A rhythm guitar region is changed, to match an arbitrary chord symbol specified by the user. The algorithm finds an optimal way to respect both the new chord and the texture of the previous bar.

These tools have been developped by Léo Dupouey and based on algorithms designed by Alexandre D'Hooge.