Ok, I opened the project with my measurements that I did some time ago and I have to correct myself: Tonex is actually worse than NAM regarding aliasing. I also re-did the measurements just to be sure I didn't screw up anything the first time, but same results.
All three are the same exact amp with gain and level matched (Fractal's VH4 captured with both NAM and Tonex)
FM9
View attachment 7418
NAM
View attachment 7419
Tonex
View attachment 7420
All three are made with a 10 seconds linear sine sweep ranging from 0 to 24 kHz at -1dBFS (
@James Freeman I know this is not a realistic level but I chose to do so just to make the lines more visible in the graphs).
How to interpret these graphs:
X axis is time in seconds, Y axis is frequency.
Brighter and yellow colors represent higher levels, faint and blue represent lower levels. The range goes from 0dBFS to -100 dBFS, so dark blue lines are not hearable in practical use.
The fundamental frequency (aka the original sweep) is the brighter yellow line which goes from the bottom-left corner to the upper-right corner or, in other words, from 0 to 24kHz in 10 seconds.
Other lines starting from the bottom-left corner are harmonics generated by the distortion of the amp.
Lines that don't start from the bottom-left corner are aliasing or... other things (we'll see later what these could be)
Let's analyse them one by one:
FM9
This is the example of how oversampling should be done properly. All the aliasing is blue coloured, so much more low in level than fundamental and harmonics. All the aliased lines are almost vertical, that means that only those relative to the highest harmonics are visible, hence having a level higher than -100 dBFS.
Also, you don't see the point where these gets reflected back at the top of the graph cuz that happens at a much higher frequency than 24 kHz, and also cuz there's a properly implemented anti-aliasing low-pass filter, this last reason is why you see all lines getting darker towards the top of the graph.
NAM
This instead can be taken as an example of how aliasing happens: the fundamental and all the harmonics get reflected exactly at 24 kHz (half the 48 kHz sample-rate) and then reflected again when they reach 0 Hz, and again and again till they reach the upper-right or lower-right corner.
All as expected here since we already knew that NAM doesn't have oversampling nor any kind of filtering.
Tonex
Compared to NAM the graph looks more chaotic. First of all we see a big yellow "V" in the middle of the graph, that's an alias relative to the 3rd harmonic which for some reason is quite high in level, maybe even higher than the harmonic itself which is quite strange, and the same happens for some other harmonics too. I don't know the reason for this but that's really bad for sure.
Then we can see the main lines getting darker at 16k, that's a low-pass filter.
But the strangest thing is that all lines have some copies above and below. These are basically the same as the fundamental and harmonics but shifted in time (both backward and forward) and at a lower level.
At first sight I didn't know what was happening and this is why I repeated the measurements today. But after thinking a bit about it I suspect this might be due to some linear phase processing inside the plugin causing (pre and post) ringing. This is just my assumption though, it could even be some by-product of the neural network used by Tonex. Anyway, if you have a more convincing explanation let me know.
Here are some facts guys, draw your own conclusions.
PS: another interesting thing I just noticed: there's also a yellow horizontal line at the bottom of all three graphs. I suppose this is DC offset since the amp generates even harmonics and therefor it has asymmetrical clipping.
PPS: or it might just be the 50-60 Hz ghost note usually called "hum"
