Glad to hear about the potential fixes. I think a bit of an extra cpu-hit is a worthy price if that is needed.So my bet is, the higher the partials, the louder interpolation noise. I think for extreme spectra with very loud partials in the top three or four octaves, we either need to use a more CPU intensive interpolation method, or we need to use a larger internal wavetable size. Both are manageable options.
When it happens without and OscFX, one can export the Wavetable and check if it's the same thing in Hive. Hive uses the same playback algorithm, pretty much.
For what it's worth, I did a comparison to Hive playback by creating a waveform with 16 sinewaves within a frame, no OscFx. In Zebralette 3 with normal -12 transpose, playing notes from the sixth octave and up results in the noise being noticable while in Hive the shape does not seem to produce it. The filters and fx in Hive were bypassed and volumes matched.
Interestingly, the sine is otherwise cleaner in Zebralette 3, whereas in Hive, a full spectrum of harmonics is very quietly there in the background. The wavetable also sounds similar to Hive in Serum for example.
Unfortunately I cannot test how this exported version of the wavetable behaves in Zebralette 3, as FL Studio crashes when I try to drag&drop the wavetable file. It seems that in this particular case, the export is a bit imprecise compared to the original wave created in Z3, so this test might have been useless. I can try again later with a different waveshape, when I have time.
Hopefully there is some relevant information here anyway.
Statistics: Posted by NeonBuddha — Mon Mar 18, 2024 10:45 am