44100 vs higher samplerates

[leondustar]

Ok, so I'm not sure whether this a bug or a feature.
I know that less-is-more, but I didn't expect it with samplerates

For some reason, I felt my sunvox WAV-exports contained slightly less high frequencies when samplerate (in preferences) is set to 48000 and higher.
I might be fooling myself here, but it really seems that during realtime sunvox-playback (F10) these high frequencies from a wav-sample ARE there
I tried comparing an original WAV-sample (from a samplepack) vs an exported WAV (played by sampler in sunvox).
These frequencies only disappear during export to WAV (I tried sampling the output using a sampler, and save that WAV, but result was the same).

Here's a frequency analysis of a hihat WAV-sample, played by the sampler-module with samplerate (in preferences) set to 44100 (A) and set to 96000 (B):

freqs.PNG (61.89 KiB) Viewed 441 times

I tried all Sampler interpolation-types of the sampler and/or 16/32bit exports, but it did not re-introduce the high frequencies.
Maybe the real test should be with a whitenoise wav?

Any idea?
Again, it could be that my ears are fooling me, in combination with a volume- or measurement-error

[AutumnCheney]

correct me if i'm wrong, but those higher frequencies at the sample rate of 44100 are actually the frequencies reflecting at the nyquist frequency (22050 hz)

at 96000, there wouldn't be any reflections until 48000 hz, hence the lack of reflections at the 22500 hz area. the lack of high frequencies at that area are probably from a brickwall filter on the sample, which presumably was encoded at 44100

here's a helpful explanation of this phenomenon: https://en.wikipedia.org/wiki/Nyquist_f ... _frequency

[leondustar]

good point. I also noticed that I didn't check what the original samplerate of the hihat was...that might also play a part since I expect the sampler also does some resampling in the background (to deal with pitchchanges etc).