How much fidelity do VST instruments really have?

[RexRed]

How about some ballpark figures?

For instance, my favorite FM synth Native Instruments FM8,

If I have a 24 bit, 96khz project, when I load FM8 will it load 24bit 96khz sine wave in its operators and carriers?

If I have a 24 bit, 96khz project and I load Overloud REmatrix will it load 24bit 96khz convolution waves?

I have some Native Instruments instruments that take a long time for waves to load.

Where is the theoretical top where the waves no longer get bigger if the project has higher stats?

48khz, 96khz, do some instruments and/or effects load 192khz files?

And if these instrument samples have an average upper limit, wouldn't I still want my projects to have a higher khz so it could meld these effects into a more pristine environment?

Forgetting file size, what level of fidelity reaches diminishing returns on average? Where do these instrument's samples usually top out at? 

And do some effects/instruments not even function right at 192khz?

How does Cakewalk handle these differences in wav khz, can different khz files exist in the same project?

Thanks for any clarification on this subject.

 

Edited July 16, 2022 by RexRed

[Promidi]

I never really thought about it.   I am too busy making music. 

[RexRed]

4 hours ago, Promidi said:

I never really thought about it.   I am too busy making music. 

Well it does not keep me up at night (much) but, I do care about the quality of my final product.

When I buy samples and instrument I do look at the khz but I need to look with a keener eye/ear now.

And when I decide which cello or piano I am going to use in a song, it is not just what sample fits best but also which sample is going to give my song the best fidelity.

I often wonder if I compensate for poor sample quality by just adding more junk into a song.

Furthermore, most people do not think about the quality of their samples and instruments because the vendors hide this info on their websites deep in the fine print if they even supply this info at all.

This is why when i try and make a qualitative judgment on which instruments I own have the best samples I currently draw a blank.

I like to think that most of them are at least 48khz. 

Another thing I have not tested, do all of my effects even work at 192khz?

I think it is quite possible they will not even be listed if they do not work at the frequency.

Is it worth it making a song at 192khz if the song will sound REALLY good when rendered down to 24 bit 96khz?

Edited July 16, 2022 by RexRed

[RexRed]

After spending some time on Google, it seems 96khz seems to be the hotspot for DAW recording.

But the question still is, are the samples and effects we own up to that level?

Are they being "upsampled" to accommodate that frequency?

[Alan Tubbs]

A lot of the pros I know use 48 /44.1.  
 

Although it is old, Lavry’s (of Lavry Engineering) white paper  showed that  the best sample rate was-in th low 60s, after everyone adapted 44.1/48/88.2/96/192K.  Higher rates have their own problems and is a waste of recording space since it ain’t better sounding.  Look him up on google.  It is pretty interesting.

But 96k is good for sample storage as opposed to recording.  Don’t lose much info as it makes it way to mp3 land.

24 bits is good, again for your DAW.  But  Make sure that the engine is running in 64 bit when mixing so the engine can take advantage of the extra bits to keep it from truncating the numbers sooner.  Older software fix etc. upsampled for the same effect.
 

there is a lot more to samples than just rates.  Always remember that first in your signal chain is the instrument and player, then the recoding chain  goes room, mic, pre amp, conversion, storage.  A flaw in an earlier step just keeps getting worse the longer you try to work around it.

I wouldn’t worry about the statistics as how much the sound works for you in context.  The best your audience is likely to hear your music on is a CD,  that is downstream of your DaAW quality

@

 

 

[abacab]

4 hours ago, RexRed said:

After spending some time on Google, it seems 96khz seems to be the hotspot for DAW recording.

But the question still is, are the samples and effects we own up to that level?

Are they being "upsampled" to accommodate that frequency?

I think that the internal sample rate of plugins is dictated strictly by the developer of the plugin. Such as the resolution of any sampled waveform libraries they supply with the product, or by DSP processing if the waveforms are algorithmically generated. You would need to discuss the logic of those individual design decisions with them.

So I'm suggesting that it's only the available audio output sample rates of the plugin that the DAW, and your audio interface can be adjusted to.

The plugin is what it is, and most of the modern ones today are capable of very high sample rates on the output side. How they get there, of where "upsampling" is employed internally in a plugin, is just a guess unless they documented it somewhere.

Edited July 16, 2022 by abacab

[Byron Dickens]

What do your ears tell you?

[Starship Krupa]

As far as sampled instruments and synths go, I know of none that adjust the rate/depth of their internal samples and waveforms based on the ones the DAW is using.

Cakewalk has built-in plug-in upsampling, which can be turned on and off for each individual plug-in (the setting is available from the menu in the far upper left corner of the plug-in properties UI, and then enabled and disabled globally in the Mix Module. I usually mix at 44, then render at 88, which has the same effect as 2X oversampling on all the plug-ins, both FX and synths.

This is based on multiple experiments I did using various combinations of Cakewalk's upsampling and render rates. The point of steeply diminishing returns (to my ears) is just 2X. So 88 covers that handily. Then for distribution, I use a converter program to generate the lower rate/depth formats.

Do some close listening tests, then go with what your ears like best. Remember: few to no listeners are going to listen as critically as you (and any mix engineer buddies). I definitely like to get the highest quality my ears can detect, but I also try to remember that I've spent most of my life training my ears to hear tiny nuances and I might be the only one who can notice them (consciously, at least).

As for the "64-bit floating point engine" option in Preferences....Cakewalk were supposedly first with that, and it is widely speculated that it had more to do with a marketing hype point than practical sound considerations. All other DAW's had to follow suit, of course, but notice how in most of them (including Cakewalk), it's turned off by default.

Edited July 20, 2022 by Starship Krupa

[abacab]

14 hours ago, Starship Krupa said:

As for the "64-bit floating point engine" option in Preferences....Cakewalk were supposedly first with that

The "64-bit floating point engine"  has to do with the floating point math that is used by the Cakewalk mixing engine, not the sample rate of the audio involved. Apples to oranges ...

[Jim Fogle]

Perhaps +++ THIS +++ David Baer article, "Oversampling - What, Why And When", just released in the July edition of Soundbyte may answer some questions regarding oversampling.

[John Vere]

I always pay attention to the file size of the sample if available. Like in Dim Pro. I assume that the larger files might be better quality but then ?  

[Quick Math]

I was thinking about these.

Let's think about a sampler. If the project sampling rate and the wave file loaded on the sampler have different sampling rate, then the sampler itself converts the data to the project sampling rate. So the quality of the SRC is determined by the sampler, and most of the time, the quality is not that good. 

So, it's probably better to use 3rd party SR converter first to get target sampling rate. And then load samples into a sampler.

But there are times where you can't do this, like Instruments where wave data are integrated within the program and users can't access to files. In that case, the degradation introduced by SRC might not be worth using higher sampling rate, perhaps.

But, then there is other thing, like sampling rate conversion from 44.1khz to 88.2 is not a complicated calculation, so the algorithm behind that might not be that different among SRC?? So, for example, SI Instruments by cakewalk by bandlab internally loads sf files and it looks these are mostly 44khz, so, for these particular Instruments, rather than choosing 96khz for project sampling rate, 88.2khz might be a better choice??

Or if that's not the case and SRC quality of an instrument is actually really bad, then sticking to 44.1khz and render it and use sampling rate conversion later for later process might be the choice.

I was testing something related on a Sampler on Bitwig Studio. Just loading 44.1khz samples on sampler in 88.2khz vs converting samples to 88.2khz by using 3rd party SRC and then loaded them on to the project. There were some sound differences and I liked the one with 88.2khz samples.

I don't know what this means... probably this means the quality for x2 conversion is still different among SRC. Maybe because of the implementation of filter...?

This is all about samplers but then what about synth plugins? They have wave forms internally as it's synth. I like Z3TA and this can load waveforms. I have some waveform collection that is 96khz 32bit. So does this synth work like a sampler? Or the waveform gets converted into mathematical formula or numbers regardless of the sampling rate as it's just a single cycle and work as it is? I don't know DSP, plugin development nor any mathematics for these. But I assume that this is how it works, so for synth, maybe whatever project sampling rate is fine? 

I might be wrong and I wanted to know how things work...

Sorry if it's a bit too irrelevant or something.

 

Edited July 22, 2022 by Quick Math

[abacab]

11 minutes ago, Quick Math said:

I might be wrong and I wanted to know how things work...

Sorry if it's a bit too irrelevant or something.

It appears that samplers and synthesizers need to be evaluated separately. It seems to me that it's a complex mix of source content, internal plugin DSP, and the project sample conversion rate for mixdown and output.

[John Vere]

Now you got me thinking and actually a bit concerned about this. I’ve been converting all my old projects over from 44.1 to 48. I wasn’t at all concerned about the midi tracks and instruments.
But one thing I’m hearing is Ample P bass lite is all of a sudden sounding grainy if you solo it. I’m going to try a listen to it at 44.1 again! 

[abacab]

5 hours ago, John Vere said:

Now you got me thinking and actually a bit concerned about this. I’ve been converting all my old projects over from 44.1 to 48. I wasn’t at all concerned about the midi tracks and instruments.
But one thing I’m hearing is Ample P bass lite is all of a sudden sounding grainy if you solo it. I’m going to try a listen to it at 44.1 again! 

It's probably best not to assume that any plugin's internal DSP is compatible with every sample rate available in a DAW, at least until you test it in your environment.

I tend to leave my DAW and interface set at 48, and so far haven't experienced any issues with that. Just fired up my Ample P lite and it sounds OK at 48.

I would expect any serious plugin issues to be compatibility with very high sampling rates.

[John Vere]

I only just noticed it as I was editing and had it soloed . First thought was my Yamaha NSM10’s were starting to die. But it was worse in the headphones.  Haven’t had a chance to figure it out and then I read this thread. 

[abacab]

24 minutes ago, John Vere said:

I only just noticed it as I was editing and had it soloed . First thought was my Yamaha NSM10’s were starting to die. But it was worse in the headphones.  Haven’t had a chance to figure it out and then I read this thread. 

Are you soloing a recorded track in your project? Maybe something got corrupted.

Open a fresh project, drop in the plug, and play it live. That's what I just did. Sounds fine at 48 on my Focusrite Scarlett 2i2.

[RexRed]

WOW! A lot of great posts here!

Tons to think about!

I usually record my projects at 24 bit 96khz but recently I recorded a song at 24bit 48khz and it sounded really good.

I wonder if there really is a benefit from not overworking the processor.

My Cakewalk graph says the processor cores are barely working but barely working is still millions, if not billions of calculations per second. ?

I have kept 64 bit oversampling on steady since it was first introduced in Cakewalk.

I read that long article and I still am not sure what it really does.

My biggest concern is that Melodyne probably relies heavily on 64 bit oversampling due to it varying the pitch of samples.

Or maybe Melodyne preforms better with it off?

Something I would like to know more about.

I usually turn oversampling to the max in all plugins right before I render.

I am wondering if it is really accomplishing what I think it is doing or making things worse?

And what if I have oversampling on in plugins and also in Cakewalk, is there an interaction factor involved here?

I question my paygrade versus my ability to understand. lol

What is the meaning of life?

Edited July 26, 2022 by RexRed

[John Vere]

People on this forum of who I trust their opinion all have stated that the 64 bit engine will do nothing for you and can possibly do harm. 

[abacab]

10 hours ago, RexRed said:

I have kept 64 bit oversampling on steady since it was first introduced in Cakewalk.

What Cakewalk setting are you referring to here?