Question on aliasing (particularly with FM)
Just got my Deluge: really cool, generally happy with it so far!
Although I love the sequencer/sampler capabilities, I quickly ran into some aliasing issues with the internal synthesizer engine (specifically with FM), and I am wondering if I am doing something wrong or if there is a way around these issues that I might have missed. (I have searched around the forums and found some posts related to aliasing issues, but this was in relation to the pitch-shifting of a multisample.)
Before buying, I had read that the sound engine is susceptible to clipping: I have noticed this as well but it is easily overcome by setting a low volume for the oscillators, and therefore not really a problem. (I also read the manual and saw that there are not too many different FM algorithms, but I actually think you can do a lot with those that are in there.) However, what I did not expect was to run into aliasing very quickly.
As a specific example, I tried to create some electric-piano-like sounds, where I'm using the modulators to 'model' the response of the tines. If I transpose the modulator up by 24 semitones (corresponding to a frequency ratio of 4, which is not too high) and I turn up the modulation depth by just a little bit, I already hear a loud alias in the higher registers of the piano (let's say from C6 and up or even C5), which sounds pretty badly distorted.
My question is then: did anybody else run into this, and does anybody know a way around this? It seems this is not easily overcome like the clipping issues. I've already tried inversely linking the modulation depth to note pitch: this works to some extend, but of course what you get then is that the lower notes become very heavily modulated. (Cool in a different way, but not what I was going for, and the aliasing still shows up pretty quickly if you're not very careful.) Is there any other setting or method hat I might have missed, such as a higher sampling rate or another way of limiting the modulation depth for high notes?
If not, I will submit it to the software suggestions (realising that it's not a very simple fix and that FM might not be that popular). The most obvious way to decrease this issue would probably be oversampling by some amount: I realise that this multiplies the CPU usage, but since you can put 8-voice unison on high polyphony, I think the CPU would be up to the task (if you don't use unison at the same time). Another way might be to limit modulation depth for higher pitch, for example by making the scale of the knob pitch-dependent or by inversely tying modulation depth to pitch but only for the higher pitches. (I think this might be what older FM synths are doing?)
Anyway, let me know your thoughts!
Edit: typo correction