sample rates
- Thread starter nznexus
- Start date
GlitchFreak
Electronic Composer
Absolutely, 44.1 Khz is a popular sample rate for CDs but 48Khz is the best sample rate mostly because that's what soundcards are generally built around. You get the distortion from oversampling simply because your soundcard doesn't know how to work with those frequency levels. I guess technology hasn't caught up to it yet..?
moses
hardliner
@nznexus: I think you are hallucinating. But please elaborate, what exactly starts to sound "unnatural and really weird"? Your productions? Or do you mean vsti's or mixes?
Actually, most AD and DA converters work substantially better at higher the sample-rates. Same is also the case for most plug-ins, samplers and synth.
Lol no. They are built to deliver their specs. And when the spec says: "Supports sample rates up to XY", it's fine.
What?!
Ughh? You know it's 2013 right? Digital audio has been around since a while in case you don't know.
Actually, most AD and DA converters work substantially better at higher the sample-rates. Same is also the case for most plug-ins, samplers and synth.
Lol no. They are built to deliver their specs. And when the spec says: "Supports sample rates up to XY", it's fine.
What?!
Ughh? You know it's 2013 right? Digital audio has been around since a while in case you don't know.
I think he's referring to aliasing. If the down-sampling process has no good LP filtering this will be the case. To be honest nowadays you can run a DAW at a massive sample rate so it's not really an issue. You're probably more restricted by RAM which forces you to use lower sample rates for content anyway, pushing it well below any nyquist limit.
moses
hardliner
I have absolutely no idea what you are talking about. He didn't mention aliasing. And as I already said, the higher the sample-rate, the easier it is for the AD or DA to do the required lowpass filtering (i.e. the higher the sample-rate the lower the aliasing will be).
GlitchFreak
Electronic Composer
Yes, I forgot to reference aliasing off the sample rate.
moses said:
Yes most standard soundcards that come with those factory built PCs still handle some crappy soundcards. I guess you haven't shopped at best buy recently? Also do your research on those sample rates to soundcards you might be interested on that sarcasm you have towards a guy who's giving his advice to someone.
Care to share with us where you got that info about sound getting distorted when oversampling? Cause I don't ever remember seeing that in Nyquist-Shannon sampling theorem.
GlitchFreak
Electronic Composer
"The rate-distortion performance, and the tradeoff between the sampling rate and the quantization accuracy is investigated, utilizing the observation that the coding scheme is equivalent to an additive noise channel. It is shown that the mean-square error of the scheme is fixed as long as the product of the sampling period and the quantizer second moment is kept constant, while for a ked distortion the coding rate generally increases when the sampling rate exceeds the Nyquist rate. Finally, as the lattice quantizer dimension becomes large, the equivalent additive noise channel of the scheme tends to be white Gaussian, and both the rate and the distortion performance become invariant to the sampling rate."
- Rate-Distortion Performance in Coding Bandlimited Sources by Sampling and Dithered Quantization
by Ram Zamir and Meir Feder
If I'm reading this wrong, by all means I am the type to admit it but from what I understand that's where distortion can be affected by the level of sample rate..?
- Rate-Distortion Performance in Coding Bandlimited Sources by Sampling and Dithered Quantization
by Ram Zamir and Meir Feder
If I'm reading this wrong, by all means I am the type to admit it but from what I understand that's where distortion can be affected by the level of sample rate..?
Last edited:
bandcoach
Zukatoku - Mod Scientist
Got a link for that?
Because what it is saying runs counter to everything I have ever read on the subject, in terms of sampling rates and encoding schemes. I think you may find that they are misusing the word distortion for the concept of redundancy, which is a well understood facet of encoding schemes.
I also note that there is no discussion of the underlying bit depth of the sampling process. The lower the bit depth, the greater the noise level in the signal, because of the factors they cite.
At higher sample rates the bit depth usually increases, which means that noise in the signal decreases proportionately.......
Because what it is saying runs counter to everything I have ever read on the subject, in terms of sampling rates and encoding schemes. I think you may find that they are misusing the word distortion for the concept of redundancy, which is a well understood facet of encoding schemes.
I also note that there is no discussion of the underlying bit depth of the sampling process. The lower the bit depth, the greater the noise level in the signal, because of the factors they cite.
At higher sample rates the bit depth usually increases, which means that noise in the signal decreases proportionately.......
GlitchFreak
Electronic Composer
bandcoach said:
Yeah here you go man, http://picasso.eng.tau.ac.il/~meir/...erformance in Coding Band-Limited Sources.pdf
Last edited:
bandcoach
Zukatoku - Mod Scientist
So this is quoted from the first section after the abstract
This says the opposite of what you said - increasing sample rate will lower actual and perceived distortion. It also pushes the discussion to the very special case of a bit depth of 1.
This whole paper seems to be concerned with compression techniques for data storage. As such the discussion, whilst relevant to audio, is in the realms of compressions codecs and as such does not directly address sample rates with specific bit depths.
When we sample we are variously using bit-depths from 16 to 32 and even 48. The higher the sample rate, the higher the bit depth, is the usual heuristic.
So, yeah sorry to burst your bubble but the paper is actually saying the opposite of what you interpreted it to mean. I'm certain that moses will have some things to point out as well.....
This says the opposite of what you said - increasing sample rate will lower actual and perceived distortion. It also pushes the discussion to the very special case of a bit depth of 1.
This whole paper seems to be concerned with compression techniques for data storage. As such the discussion, whilst relevant to audio, is in the realms of compressions codecs and as such does not directly address sample rates with specific bit depths.
When we sample we are variously using bit-depths from 16 to 32 and even 48. The higher the sample rate, the higher the bit depth, is the usual heuristic.
So, yeah sorry to burst your bubble but the paper is actually saying the opposite of what you interpreted it to mean. I'm certain that moses will have some things to point out as well.....
Last edited:
GlitchFreak
Electronic Composer
If there was any part of a sarcastic point someone was trying to make and I cared for that view, I'd be concerned about what moses had to say. As far as I'm concerned, he should keep his views to the liberation of Egypt. Thanks for pointing out my misinterpretation I'll be sure to go into further investigation of a synopsis made in a paper.
It seems you've misinterpreted the term "rate-distortion" as a trade off between sampling rate and distortion. You can read what "rate-distortion" is all about by clicking here.
The sampling theorem states that the signal can be faithfully reconstructed as long as the sampling rate is two times the bandwidth of the original. That is why aliasing cannot occur in 44.1kHz or higher sample rate since the audio signal is filtered above ~22kHz with a LPF (although not ideal LPF).
The sampling theorem states that the signal can be faithfully reconstructed as long as the sampling rate is two times the bandwidth of the original. That is why aliasing cannot occur in 44.1kHz or higher sample rate since the audio signal is filtered above ~22kHz with a LPF (although not ideal LPF).
bandcoach
Zukatoku - Mod Scientist
Moses actually knows quite a lot about this (he is a mod here and mods here only to get to become one if they demonstrate superior knowledge and application in the fields of audio production) and whilst he may come across as sarcastic, it is more a case of the grammatical structure of his native language (German) than any intentional attempts to have ago at you or anyone else.
so peace out....
The only way a sample used at a higher sample rate than the DAW is in the instance that the DAW down-samples without a proper LP filter.
The paper is about a theoretic situation in which you can get a better replication of a source by increasing the sample rate while decreasing the bit depth. This achieves lower distortion (distortion technically is just any unwanted component to the signal). If you use an exceptionally high sample rate...I mean crazy high, you can use a bit depth of 1 bit. Which is used in some audio formats. Its called sigma-delta. But it uses a bit of a hack since the samples aren't values like on a CD they are 'deltas'.. which means that they are the difference between the current and previous value. Up 1...Up 1...Down 1...etc etc. This type of coding has the advantage that it requires very little memory as you only deal with 1 bit at a time. You just need a crazy clock speed. Which nowadays is doable. Some argue this is a far better use of available storage space. Actually I think it does actually cause a lot of distortion (a continuous sawtooth) but it is at such a high frequency (given the gigantic sample rate) that it is far beyond audible.
The paper is really about an analysis of the situation in between the two extremes and what the trade offs are.
Anyway, none of this applies to any DAW you are likely to use as the sample rate is fixed and the bit depth is likely floating point.
The paper is about a theoretic situation in which you can get a better replication of a source by increasing the sample rate while decreasing the bit depth. This achieves lower distortion (distortion technically is just any unwanted component to the signal). If you use an exceptionally high sample rate...I mean crazy high, you can use a bit depth of 1 bit. Which is used in some audio formats. Its called sigma-delta. But it uses a bit of a hack since the samples aren't values like on a CD they are 'deltas'.. which means that they are the difference between the current and previous value. Up 1...Up 1...Down 1...etc etc. This type of coding has the advantage that it requires very little memory as you only deal with 1 bit at a time. You just need a crazy clock speed. Which nowadays is doable. Some argue this is a far better use of available storage space. Actually I think it does actually cause a lot of distortion (a continuous sawtooth) but it is at such a high frequency (given the gigantic sample rate) that it is far beyond audible.
The paper is really about an analysis of the situation in between the two extremes and what the trade offs are.
Anyway, none of this applies to any DAW you are likely to use as the sample rate is fixed and the bit depth is likely floating point.
Last edited: