After creating a rendering of the 1st aria in Bach's BWV 170 I noticed a nasty 'glitch' in the Oboe d'Amore part where a C4 was being held for a relatively long time. I played it on the stand alone Aria player and sure enough both the C4 and the B below glitched.
Investigating the appropriate sfz file I find that they use the same sample.
"<region> lovel=1 hivel=127 lokey=59 hikey=60 pitch_keycenter=60 amplitude=100 tune=0 loop_mode=loop_continuous loop_start=140796 loop_end=147402 sample=OboeDAmorC4_0001102C.audio"
Opening the C4 sample (OboeDAmorC4_0001102C.audio) in Audacity and entering the start and end points of the loop, I find that the start point (140796) is at a zero crossing point but the end point (147402) is at a wave peak. Experimenting with changes to the end point, I was able to get a zero crossover point and the correct position within the overall wave form, at 147388.
Editing the sfz file to reflect the new loop crossover end point and reloading the sample has effected a complete cure.
Should I post the corrected sfz file?