creart wrote:When comparing sounds 'out of the box' I couldn't say Pianoteq was the best, but for instance the full velocity range instead of sample-switching in a couple of steps over the velocity range seemed a better option to me.
On that part I wonder what sample libraries will benefit from keyboards with much more velocity steps....
cheers
Hans
When I compared sounds, I did several things;
1. Downloaded demos songs produced by the sample manufacturer and posted on his website. I would assume that they've put forth their best effort to make their product look good. I concentrated on classical piano because midi files for these are easy to find, and often are available in several versions.
2. Obtained the same songs (the midi files) and rendered them in Pianoteq to wave files.
3. Using a wave editor, I loaded a number of songs, and compared them carefully by switching from song to song.
Invariably, the results were the same:
1. Some notes of the sample created files went "boonk". That's about as accurate as I can describe it verbally. Note that the "boonk" sound was in the bass, where samples might theoretically be better because they are actual recordings, and bass notes are much more difficult for render than high notes (this applies to PT).
2. In general the songs rendered with samples had little or no sustain. This is interesting because they claim to have used concert grand pianos which have wonderful sustain.
3. In general the songs rendered with samples displayed little if any Sympathetic Resonance, something all acoustic pianos have, and in the case of a concert grand, the SR is wonderful and rich.
4. Even the supposedly best sample (Ivory) did not have the clarity that the PT rendered piece did. By clarity I mean that individual notes can be heard in a chord or complex passage. The sample based songs are seldom as clear.
Now to address your comment in regard to multiple velocities. Pianoteq already does this; each note is generated with the velocity associated with the note.
The best samples (as far as I know) have three velocity layers. As we know, the harmonic output of an acoustic piano varies with velocity (harsher at high velocities, mellower at lower velocities). In order to accurately duplicate this tonal range, a sample library would need many more levels of velocity sampling. Keep in mind that midi has 128 levels for every parameter including velocity.
Sample libraries already are huge - I don't know how many velocity levels would suffice, but in theory they should have 128 levels. Perhaps only 100 would suffice, but if they are gigantic with 3 levels, please don't try to find a computer that could handle 100 levels. Perhaps Big Blue, the chess champion, could do it, but it's currently beyond my budget capabilities.
I've posted this site before, but it's certainly worth listening to:
http://www.purgatorycreek.com/
When I downloaded the midi file for this "musical selection" and rendered it in Pianoteq, mine sounded better than the one for Pianoteq on the site. I'm wondering if it was done with an older version of Pianoteq.
In summary, when I compared the sounds "out of the box", I literally laughed out loud at some of the sample based sounds, the rest were just bad.
If you have found a better sound, please let me know what created it.
Glenn
__________________________
Procrastination Week has been postponed. Again.