Currently the algorithm for stitching together bits of bass phrases to a progression can be jumpy melody-wise and very obviously "BIAB" when compared to - for example how a real pro double bassist like Ron Carter would approach a line.
And Ron Carter IS one of the artists with real tracks here!

The problem is that sometimes there are big jumps interval-wise where there is a chord transition. A real bassist will usually have grace notes and lead-in notes that are close to the upcoming strong chord-tone.

But the current algorithm doesn't take note of this. It might be that when the ability in BIAB to have per-note pitch transformation possible this might be achieved. ( I actually did this myself with a version of the zPlane software on my own project may years ago to solve similar issues ).

But until that - it just requires a more judicious choice of "snippets" motifs and lines that sound more musical and realistic in a walking bass context.

Because - for example - currently on YouTube - where many have uploaded jazz backing tracks for people to use - it is very very obvious when the jazz backing track came from BIAB.

Even with Ron Carter on the real tracks.. it gives the game away that it ISNT Ron Carter... because the lines just aren't ones one hears Ron doing. on any of his real material using standards.


As it happens I don't find this an issue so much with non-walking bass parts for non-jazz styles and that is because those styles of music tend to often have the bassist doing motifs in a manner more suitable to the BIAB algorithm.

But jazz walking bass is where the inauthenticity becomes obvious. At least it does to those who play jazz or have listened to it as a fan for many decades.

Big Sur, Mac Mini i7 2018, 64GB RAM 2TB SSD
Yamaha PSR SX 900, Cubase, Ableton, UNIFY, OPUS ORCHESTRATOR, MASCHINE, Vienna Ensemble Pro ,Audiofuse, L6 HELI, KOMPLETE, Komplete Kontrol 88
Lockdown beard, podgy face, beer belly, bald patch ... yadda yadda