I’ve been contemplating this question lately and was wondering what the developers of SWAM instruments thought about it.
To me, physically modelled virtual instruments such as those by the SWAM and Sample Modeling teams bring a whole new level of expression and aliveness to the world of virtual instruments. With the aid of good MIDI controllers, the ease of playability makes it incredibly easy to infuse a natural, subtle and very intuitive expression in them. It’s a totally different approach to sample-based instruments, which in comparison demand a much more analytical and awkward process : you think about the sound result you would want from an imaginary instrument (ie : the “real acoustic” violin) and then see if there’s an actual sample that matches what you have in mind. Working with physically modelled instruments, on the other hand, asks you to play the instrument and see what good musical results you can get out of that dynamic exploration. I think this is a much more fun and fruitful interaction with the instruments. It's an approach that’s very much in line with my view of how virtual instruments should be treated : with a high level of technical mastery for a more intuitive, natural and subtle expression. This is why I think these sort of instruments are the future of virtual instruments, especially sustained instruments (I think it’s a bit different with percussive instruments…).
So on the dynamic level (behavior-wise : transitions, legato, expression, etc…), physically modelled instruments are in a whole other league, and this, to me, is probably the most important aspect in terms of musicality. However, on the static level (timbre, sound quality, richness, frequencies, etc…), I think it’s fair to say that there’s still a bit of catching up to do with sample-based instruments.
I’ve been playing with SWAM woodwinds and string instruments for about a year now. At first, I was so thrilled that I practically envisioned using only SWAM instruments and Sample Modeling’s brass for all non-percussive instruments, preferring to focus on these sort of instruments to become as efficient as I could with them. However, it soon became obvious that I was starting to miss the richness of samples in some circumstances. This is true especially of instruments which typically produce a high level of noise and artifacts, such as strings and flutes. After some time, I started to realize that although on the level of expression, the potential was limitless, on the level of sound itself (again…all the small nuances that makes an instrument sound rich such as the air/breath in the flute or the bow noise in the strings), it sounded a bit too much of the same. It lacked small errors, squeaks, artifacts…all those little details that make the sound feel alive and natural. Today, therefore, my approach is to make the best of both worlds : I use SWAM and Sample Modelling instruments for any passage that’s legato or with a lot of transitions, and I use samples for long single notes, extended techniques, percussive sounds, etc…this helps to add some richness to the overall sound.
So this is where my question comes in. Do you think this is an inherent limit to the art of physical modelling or do you think this is something that is to be perfected over the years? I know nothing about the technology behind it, only that it’s still in its early years. Can synthesis ever TRULY create a static sound that will be as rich and beautiful as one long violin note? Can it realistically reproduce all those tiny variations in the sound waves. Or is this simply not its destiny? I use the violin example because, to be honest, the strings are probably the instruments that bother me the most, although bother might not be the best word because I’m still blown away by the amount of expression I can put into them. However, I think it’s fair to say that if you play one single note (without dynamics) on the violin to anyone and then compare that with a sample-based instrument, anyone will prefer the sound of the sample-based instrument. There is still something a bit too synthetic/artificial about the strings’ timbre. They only ever become natural when you start adding a lot of expression in them ; then they become alive. But sometimes, what you want is just a simple, very soft, long note that fades out to an almost inaudible sound, leaving only that beautiful bow noise…there, you have really come out of the SWAM Strings' comfort zone. So is this, as an example, something that would be possible to achieve over the years? Or is it just not a kind of behaviour or sound that can be reproduced? If it IS possible, how close are we to actually seeing this kind of improvement?
I ask this because I’m very curious about what the future might hold in terms of virtual instruments. Are we moving in a more horizontal direction, extending the number of instruments created with this form of technology until a new and better form of technology arises and brings a new level of realism/quality, or are we also going to move in a vertical direction and deepen the level of quality and realism of instruments created through physical modelling? I can see from SWAM’s future development thread that you guys seem to be focusing on expanding the range of instruments with the new ensembles project and the world instruments (also very exciting!), so do you feel that the instruments you have crafted will remain more or less the same for the next few years?
In all cases, I do think the future is bright and I’m grateful for the amazing work you guys have done to provide us with wonderful expressive instruments!