Page 2 of 2

Re: internal sampling rate

Posted: Tue Jul 07, 2020 12:48 pm
by HowlingMod
i use, almost from the beginning, 24 bit/48Khz, no real big decision, or reason. nice compromise.

the internal sample rate of 48Khz of VM seems to be necessary to constrain (...) anti-aliasing a bit, and i think it is also for audio rate. it must be fixed.

in reaktor you can change the internal sample rate, and more plugins use internally other sample rates, or do oversampling.

there is not much difference in quality between 48Khz and 44.1Khz, but i prefer it. i am used to it.

the resampling is the task of a plugin.

in reaktor you can indeed here the difference between 48Khz and 96Khz, internal setting. not always that big. i let it stay at 48Khz. it is one core plugin, even a 5Ghz AMD or Intel, will.. o well, i also never change the buffer size, i stand by my buffer size! that can help. o well when you set your project for instance to 96Khz, double the buffer size... (latency will be the same..) somewhat confusing answer. haha.

no option for offline rendering, but perhaps that is choice, for reaktor, because it really changes the sound.

but internally VM is always busy, when things are connected, at a fixed sample rate. i know the basics, and sometimes my head works better..

for your own projects, your daw, 44.1 or 48Khz does not change much.

although 48Khz has a small advantage for nyquist.

that is why, i think cherry audio has choosen for this sample rate, 96Khz would put a lot of strain on the cpu (and not only the cpu, it is complex, real time audio within the realm of a computer).

why the bit depth of 24 bit in my case, bit depth of my interface (but in reality only 20 bits are used...). more headroom.

EDIT: it seems that VM uses a low-pass filter for Nyquist also, only sample rate doesn't do it. there are more ways to prevent anti-aliasing.
i don't much about filters for nyquist. chopping of frequencies is one way, but it must be done before...

oversampling is another way, but it puts strain on the pc.

see the other reactions in this thread.

Re: internal sampling rate

Posted: Tue Jul 07, 2020 4:31 pm
by andro
Still having trouble seeing where the signal is coming from. Pardon my ignorance. What is generating that frequency spike on the chart?

Re: internal sampling rate

Posted: Thu Jul 09, 2020 5:45 pm
by Harry Mudd
hi andro,
Plugin Doctor (PD) is an audio test enviroment with a generator and several audio analysis tools.
The test device VM is fed with the generator of PD with a test signal suitable for the selected measurement.
i.e. for a frequency response with a sweep siganl, for a harmonic distortion measure with a user selected sine wave etc.
What you see in the graph I posted is the PD generated signal for the harmonic distortion meaurement sent to VM with no modules inserted direct to the output of VM then back to the analysis part of PD. Normally the signal should look like the 48kHz plot. But due to internal behaviour of VM the 44k1 signal is distorted by the resampling process in VM shown in the corresponding plot.
Hope this helps.
H

Re: internal sampling rate

Posted: Thu Jul 09, 2020 6:26 pm
by wavemechanic
My point is that the sample rate converter in a 96kHz DAW ought to have a steeper curve (from 19-22kHz). Instead it's grabbing even below 10kHz, dulling the sound. Personally, I think VM would sound better if it natively supported 96kHz as well, since triangle, square and any distorted waveforms definitely cause audible aliasing to bounce back off the Nyquist frequency. If we are going for analog realism, 48kHz doesn't really cut it. It's fine for a sine wave but not a square wave, unless VM is oversampling at multiples of 48k.

Re: internal sampling rate

Posted: Fri Jul 10, 2020 10:13 pm
by cherryaudio Greg
Hi,

We released 2.0.25 today, and it has improved resampling for sample rates lower than 48000 Hz.

Give it a try!

Greg

Re: internal sampling rate

Posted: Fri Jul 10, 2020 10:25 pm
by wavemechanic
Hopeful Request (fingers crossed and holding breath): Really hope the SRC will be improved for people working at 96kHz soon. That SRC filter for 96 is poor, to put it kindly. I'm praying for a filter that starts at 19kHz to -144db at 22,050 kHz or something similar. Basically, I just want the filter not to butcher 10-19kHz.

Futile Request: I wish we had the option of running natively at 96kHz. Aliasing at 48kHz from harmonic content is not insignificant and well within the audible range as seen in this image of a square wave from VM's basic oscillator.