Thanks for the tip. Yes, this issue is occurring at the joins at the end of the measure prior to a "B" partmarker. (or an "A" partmarker on a middle chorus configured to "vary style on middle choruses" as is the case with my examples).

I've also discovered that the amplitude and "distinctiveness" of these artifacts increase as the tempo of generation increases. For example, this same Jazz#1 RD generated at 120bpm, while it still shows evidence of this artifact, isn't as amplified and is more diffused (resulting in it being less obvious) that if it is generated at 250bpm - which is very obvious. Unfortunately, generating slower and then setting for much faster playback isn't an good strategy as the Jazz #1 RD generating at 120bpm doesn't sound realistic played back at 250bpm. So, the clean-up technique you demonstrate above is the the only option until this is addressed by PGMusic.

SR

BIAB/RealBand 2025 Audiophile | GPO 5 | Windows 11 | Martin TR4501