Quick Answer: Multiple problems can lead to this. Read more below.
Track type analysis in Liquid NotesWhen opening a project, all tracks of your MIDI arrangement are classified according to their respective musical function such as melody, chords, bass, or percussion. This is done by facilitating state-of-the-art machine learning methods that look for semantics used in the name of instruments and patterns in notes for each track to determine the type of track.
The result of this analysis is shown to you in the Open Project Wizard, where you can make changes to the classification suggested by Liquid Notes before performing the harmonic analysis.
Wrong classification of tracksIt can happen that the track type classification delivers wrong results. This is caused by it falsely detecting names associated with melody, chords, bass, or percussion in the name you assigned the tracks (instruments) in your arrangement, or by the algorithms not being able to clearly determine what kind of track it is looking at from the note information (MIDI events) derived.
While naming of a track is a simple issue, being able to classify a track has some blurry lines: for example, if your entire arrangement consisting of different instruments is 'garbled' into one track only, Liquid Notes will display an error message in most cases. Should this merging of instruments not be detected by the algorithms, it may lead to a wrong classification of your track.
With music being a broad field with near-endless opportunity for variation, we constantly work on improving our algorithms and the above are only a few examples why tracks may not be classified correctly.
Send us your arrangement if you continue to receive wrong track classification results!