Maybe a stupid question, what is the sample rate of hardware synths?

I just thought about this, digital hardware synths must have some set sample rates (I assume analog ones as well?) but no site really has any details on this.

Are they in general 44.1khz, or does it vary from synth to synth? And how is this different in analog synths? Is there basically no sample rate, or does it depend on whether or not certain parts of the synth are digitally controlled, etc?

analog synths don't have any sampling rate because there is nothing being sampled, it's all continuous voltage instead of discrete samples/snapshots of a signal at 44100 times a second or whatever. sampling rate is only relevant when there is something being sampled (analog to digital).

  iep said:

sampling rate is only relevant when there is something being sampled (analog to digital).

Not true, it matters wherever there's DSP, including digital hardware synths (digital to analogue). Don't know what internal rate is used, but 44.1khz is a good bet for most.

One way to look at things is that an analogue signal has an infinite sampling rate. I'm not sure what you mean by digitally controlled. If you're referring to DCOs its the tuning that is controlled, the signal path is still analogue.

Edit: The original korg wavestation was 32khz. Probably the same for most synths from the period.

  kakapo said:

  iep said:

sampling rate is only relevant when there is something being sampled (analog to digital).

 

Not true, it matters wherever there's DSP, including digital hardware synths (digital to analogue). Don't know what internal rate is used, but 44.1khz is a good bet for most.

usually you don't say "sampling rate" for internal rates because there is nothing being sampled. most digital synths work multirate anyway, high rates for operations that require precision and lower rates for operations that require speed.

  iep said:

  kakapo said:

  iep said:

sampling rate is only relevant when there is something being sampled (analog to digital).

 

Not true, it matters wherever there's DSP, including digital hardware synths (digital to analogue). Don't know what internal rate is used, but 44.1khz is a good bet for most.

 

usually you don't say "sampling rate" for internal rates because there is nothing being sampled. most digital synths work multirate anyway, high rates for operations that require precision and lower rates for operations that require speed.

'Sample rate' is acceptable terminology, and you do usually say sampling rate when talking about it.

Yeah, that's actually why I was scared people would say it's a stupid question, because sample rate is usually used for things that play recorded sound, or anything having to do with an audio sample... Though even then, I don't really see how a digital sound processor can make a sound without having some sort of sample, even if it's just basic waveforms. I just never really realized how little I actually understand about how digital synths work, and even a fundemental detail such as what sample rate they may use was completely overlooked by me until a few days ago when I thought about it...

I wonder if analog modelling synthesizers use a higher sample rate, to get closer to the analog sound, but who knows really.

I bet there are ways to find out the sample rates of various synths, but it'd probably require an electrician checking the various chips with an oscilliscope but even then who knows.

  Tamas said:

Yeah, that's actually why I was scared people would say it's a stupid question, because sample rate is usually used for things that play recorded sound, or anything having to do with an audio sample... Though even then, I don't really see how a digital sound processor can make a sound without having some sort of sample, even if it's just basic waveforms. I just never really realized how little I actually understand about how digital synths work, and even a fundemental detail such as what sample rate they may use was completely overlooked by me until a few days ago when I thought about it...

 

I wonder if analog modelling synthesizers use a higher sample rate, to get closer to the analog sound, but who knows really.

 

I bet there are ways to find out the sample rates of various synths, but it'd probably require an electrician checking the various chips with an oscilliscope but even then who knows.

iep is right in one sense, in that the truest definition of 'sampling' is the conversion of an analogue signal to a digital one.

All of this also applies to soft synths. You're right in thinking that analoque modelling sometimes uses higher sampling rates to get closer to an 'analogue' sound (primarily to get rid of what's known as 'aliasing', ). If you want to hear what all this in effect means, get hold of a copy of Reaktor (demo, warez, whatever). On the front panel it has a 'sample rate' option which will take you all the way from 11khz to 192khz if I remember correctly.

  iep said:

racist

it's also very important for the frequency range: with a sampling rate of 44.1khz, the highest frequency you can get is 22.05khz (half the sampling rate, the nyquist frequency). that Korg Wavestation was a sample-based synth, it used samples that were sampled with a 32khz sampling rate, so the highest frequency of the samples was 16khz (high enough for most people). if it were a really clever synth it would've used different sampling rates for different samples, so that a harmonically simple sample would be sampled with a low rate and a complex one with a higher rate.

  kakapo said:

iep is right in one sense, in that the truest definition of 'sampling' is the conversion of an analogue signal to a digital one.

He's not right "in one sense", he's just right. There is no "sampling" if you aren't converting Analog to Digital. It's not just the "truest" it's the definition being asked about here. If you have a purely analog synth, there is no sampling at all, all other types have sampling rates that vary based on model and you can look them up. That's about it.

Yeah I've been working on finishing my newest project, and while most of my previous works have been recorded at 44.1khz, I've recently realized that the higher frequency range sounds way better, so I've started mastering my tracks at 96khz (and higher if possible), and it's weird, even when they're sampled down it still sounds way better than something that was originally recorded at 44.1khz.

I've only seriously started thinking about how important the sample rate is in the music I make solely on my laptop a few months ago (which is funny since I've been making music strictly on the laptop for something like 2 years now), and I mean, I did know that it makes a noticeable difference, but I suppose I didn't pay much attention to it. But yeah, it got me thinking about hardware as well, and how it may possibly effect my choices on what types of synths I'd want to purchase in the future.

I just wish I knew what the hardware worked at. I know for sure, that the sounds that my MS2000B makes definitely sound a lot crisper than a VST running at 44.1khz.

tftt: do you know where they have info about the sample rates of various synths? I looked for at least 20 mins on Google and found nothing, maybe I just suck at Google.

  Tamas said:

I just wish I knew what the hardware worked at. I know for sure, that the sounds that my MS2000B makes definitely sound a lot crisper than a VST running at 44.1khz.

I would bet that has more to do with the algorithms used inside of your VST than anything else. The differences among sampling rates hiher than 44.1khz are very very slightly nuanced to the human ear.

  Tamas said:

tftt: do you know where they have info about the sample rates of various synths? I looked for at least 20 mins on Google and found nothing, maybe I just suck at Google.

Well it depends, some people advertise them I don't pay much attention though. Sometimes they'll be in the user manual and you can find them in the tech spec pdf's on the manufacturers websites, others you can't find at all.

  Tamas said:

Yeah I've been working on finishing my newest project, and while most of my previous works have been recorded at 44.1khz, I've recently realized that the higher frequency range sounds way better, so I've started mastering my tracks at 96khz (and higher if possible), and it's weird, even when they're sampled down it still sounds way better than something that was originally recorded at 44.1khz.

 

I've only seriously started thinking about how important the sample rate is in the music I make solely on my laptop a few months ago (which is funny since I've been making music strictly on the laptop for something like 2 years now), and I mean, I did know that it makes a noticeable difference, but I suppose I didn't pay much attention to it. But yeah, it got me thinking about hardware as well, and how it may possibly effect my choices on what types of synths I'd want to purchase in the future.

 

I just wish I knew what the hardware worked at. I know for sure, that the sounds that my MS2000B makes definitely sound a lot crisper than a VST running at 44.1khz.

 

tftt: do you know where they have info about the sample rates of various synths? I looked for at least 20 mins on Google and found nothing, maybe I just suck at Google.

you're right - older digital hardware synths convert their signals to analogue at lower quality than newer ones.

e.g. a mark 1 DX7 only has a 12 bit digital-analogue converter, so it sounds "grungier" than a mark 2

edit: unless the instrument has a digital output, it's an analogue instrument as far as your recording setup is concerned. you dont neet to "match the rates" or nything like that - merely be aware that more D/A conversions than necessary are happening.

  ten fingers ten toes said:

  kakapo said:

iep is right in one sense, in that the truest definition of 'sampling' is the conversion of an analogue signal to a digital one.

 

He's not right "in one sense", he's just right. There is no "sampling" if you aren't converting Analog to Digital. It's not just the "truest" it's the definition being asked about here. If you have a purely analog synth, there is no sampling at all, all other types have sampling rates that vary based on model and you can look them up. That's about it.

I think you're missing what kakapo meant here. IEP's original answer was inaccurate and misleading in the context of what the op was actually asking.

People use the term 'sampling' in more than one context, considering he was disputing his answer, kakapo felt it was fair to clarify what was actually accurate about IEP's answer.

Everything kakapo has said has been clear and accurate. You, sir, are an awful, awful cunt.

in one corner! the perfectly groomed kakapo and his devoted dupey lady! in the other! the amiable kofi annan of watmm tftt and electronic music viceroy iep!

"give me one example of a sampling rate inside of a DSP chip aside from the A/D stage and the other obviously sample-based sections like in kakapo's Wavestation? plus maybe a very good reason to call that sampling rate instead of the more conventional rate?"

  iep said:

his devoted dupey lady! !

How dare you.

What has kakapo said in this thread that is inaccurate, in the context of what the op was actually asking? Open up Reaktor, why is the option to set the internal 'sample rate' of Reaktor labelled very clearly as 'sample rate'? Yes there may be no 'sampling' in the sense of A/D conversion, however it is entirely reasonable and convention to refer to the 'internal sampling rate' or 'sample rate' of a digital synth. I agree, something like 'internal processing rate' is probably more accurate when talking about DSP. I'll state it again, was the op asking about an A/D stage in hardware synths, or was he asking about the 'internal processing rate'? Is it reasonable to refer to 'sample rate' in place of 'internal processing rate' when there are digital synths that use the terminology of 'sample rate' to refer to this? I think it is. Kakapo is well aware of what you are getting at, which is why he included, "iep is right in one sense, in that the truest definition of 'sampling' is the conversion of an analogue signal to a digital one."

I believe one of the reasons 'sample rate' is used, while not technically perfect, is because it avoids any confusion with the concept of 'internal resolution', i.e. bit depth.

i'll make an exception & give a shit because there's already so many people giving each other shit

this reaktor setting now, what does it set? the rate at which the oscillator's wavetables are read? the rate at which the filter coefficients are updated? no, it tells your soundcard's driver what to set the sampling rate of it's input and output stages to (a/d and d/a). well, more technically, it allocates less or more space in the little ram buffer stage just in front of or behind the input and output stages and instructs the processors to read it slower or quicker, usually interpolating when slowing down, and then tells the actual a/d or d/a chip to read that buffer at a specific speed.

as your handsome partner said before, you can easily hear an example of if you lower the sampling rate to say 8khz and try to play a sound with harmonics above that 8khz, the high frequencies will get wrapped around nyquist and cause aliasing. but what do i mean with sampling rate in terms of d/a conversion then? well, the d/a will work as a sort of inverse sample-and-hold that samples at a clock rate (sampling rate) and then smoothens it out with a simple low pass filter to reduce the typical sample-and-hold steppiness. i say inverse, because the output is the smooth signal and the input is discrete and stepped. i'll just group them together as 'a/d' or 'output stages' or whatever because it's basically the same as an a/d, but inverted, and if you look at the guts of a soundcard you'll notice that often the a/d and d/a share a single chip.

now then this doesn't have anything to do with internal processing rates, it's just the time-based resolution (instead of the frequency based resolution that you can set with bit depth / word length) of the sample-and-hold device. it won't make your oscillators or other elements do anything slower or faster. it's just a measure of i/o resolution.

if you'd look in the source code of reaktor you'll see for example that the lfo modules would run at a much lower speed than the regular oscillators to save cpu cycles. it wouldn't make any sense to use the same processing rate for every element anyways, that would be horrible coding. sampling rate's just global i/o rate.

where's me filter cause i believe i need some smoothening out myself, this thread has me feeling a bit steppy.

we're only in it for the electronic music points, watch em rack up as topic starter keeps wondering wtf

Signed,

Yours internally,

- Mr. I. E. "i'm here to rate you" Pedantic

Some of what you say is certainly correct, and I'm not disputing that, but can we please just agree that tftt is an awful, awful cunt. I'm going to get back to you. In the meantime a quote from an old reaktor review,

"The actual sound quality of the software is quite good, and it has choices of several sampling rates in which to do its calculations, regardless of the output sampling rate."

Maybe he's making the same mistake that I am.

  Lady kakapo said:

but can we please just agree that tftt is an awful, awful cunt.

Wait, what? I'm referring to the DAC in hardware synths when I say internal sample rate. When I quoted 32khz for the wavestation, that's what i was referring to, I have no idea about the actual samples used.

I'm not 100% sure what Lady kakapo is talking about, but I'll have a word and get back to you.

  Lady kakapo said:

Some of what you say is certainly correct, and I'm not disputing that, but can we please just agree that tftt is an awful, awful cunt. I'm going to get back to you. In the meantime a quote from an old reaktor review,

 

"The actual sound quality of the software is quite good, and it has choices of several sampling rates in which to do its calculations, regardless of the output sampling rate."

 

Maybe he's making the same mistake that I am.

can i get you a drink?

  LUDD said:

can i get you a drink?

  Lady Kakapo said:

Buy me a 3l bottle of white lightning, and I'll totally lick the tip of your penis sticking up under your belt.

This thread is a veritable goldmine of classic EKT pedantry. DSP based synths, regardless of their structure and regardless of whether or not they have an analog output, all convert their signals into discrete-time at a specific rate before they are fed to the outputs. Regardless of the mystical "internal sampling rates" of the various blocks within the voice architecture, there is indeed a final "sampling rate" that determines the bandwidth of the final signal.. Which is probably what this poor sap Tamas was originally asking about.

To answer the question, different DSP-based synths process internally at different rates and deliver their final output at different rates. I believe the Access Virus and the Nord Lead are both 44.1kHz. The Nord Lead 2X and the G2 are both 96kHz. I'd be surprised if your MS2000 wasn't also 44.1. It might have a more brilliant sound due to less aggressive anti-aliasing than your used to? I dunno..

As an aside, I always wished I could halve the polyphony on my Virus and double the sampling rate to hopefully smooth out the top end a bit...

  iep said:

(instead of the frequency based resolution that you can set with bit depth / word length)

wouldn't that be amplitude-based resolution? :shade: