Thanks. Using only the mf samples, octave-doubled the basses from about b3 on down (adding just a tiny bit of 1Hz lfo), made two velocity layers, the upper with no filtering, the lower with a touch of lowpass. Applied a little Pianoverb to the finished dry rendering to mimic the lute\'s \"mulled\" sound (I might use more if I were emulating a baroque lute), then EQ\'d to thin it out a bit.
It\'s still not that lovely luminous Paul O\'Dette sound, in which the attacks sound more \"rubbed\" than plucked, but I\'m going to start fiddling with the raw samples next, to see if I can achieve that.
Just more a note of support here versus technical observations as I appreciate your motivation to work with some of these beautiful lute works. I play the lute (not well) and also sequence with the Early Patches lute samples, you are doing a fine job with Xsample variant.
In regards to troubleshooting, I suppose I would reduce the volume of the bass notes a bit, especially the d, and try to get a few slightly more dynamic fluctuations in a few of the passages with repeated figures, realizing that this aspect has only limited variability in reality.
You certainly have excellent phrasing, the arpeggio rolls are very nice. The rubbed analogy for O\'Dette’s playing is a good one. (However, in my opinion I think his choice of tempos were unfortunate with his renditions of the famous Ancient Airs and Dances heralded by Respighi).
Keep up the great work, I wasn’t familiar with that Dowland piece and enjoyed it a lot.
Thanks for the encouraging comments. In my XSample patch the basses are progressively attenuated the lower they go, up to about 10 db as compared to the trebles; but maybe that\'s still not enough. I certainly did notice that they were unnaturally loud in the final mix, and attributed it to Pianoverb resonances, but left it that way afraid I might not win over as many uninitiated listeners if the basses were more naturally weak. Think of it as a theorbo with 8-foot-long bass strings ;-) It doesn\'t sound quite so unnatural through headphones.
I have the Early Patches lute and the Bolder ones as well, and hope eventually to get the sort of sound I want from them, after messing about some more with the raw samples.
You\'re right about the dynamic variations. The piece was played-in realtime, but of course the chord rolls, the one hammer-on/pull-off trill, and a few obviously needed velocity tweaks were done afterwards. I could and should have messed around more with the velocities, but didn\'t.
The mockup exercise has caused me to think hard about the limitations of MIDI velocity as a real-time control: every trill by necessity has high velocities, for example, and fast runs are almost as problematic. Although I\'ve had limited experience with real pianos, I can imagine that even a real piano is better than a MIDI keyboard about responding to the amount of \"work\" put into the key, as opposed to the speed of the keystroke, so that perhaps you can come a lot closer to playing soft, fast passages on a piano than on a MIDI board. I can think of one or two well-known lutenists who do in fact sound as if they lose control when playing fast and consequently always play fast passages loud and punchy (and no, I\'m not dissing Bream, whom I adore), but for me the beauty of using the alternating thumb-forefinger technique on a lute or guitar, even for a non-purist like myself, is that I can control the dynamics of fast passages -- even trills -- ever so much better than I can with traditional classical guitar technique. But it seems to me that playing in with a MIDI board will always necessitate a certain amount of after-the-fact tweaking.