44.1kHz vs. 88.2kHz

[ze2be]

Ive done some research about how they make vinyl. They make the print master disk with a needle. This needle draws the wave from the master file. Resolution and sample rates above 24/44.1might not be such a difference, since the wave is transfered with an "analog" mechanical arm to this needle. And the needle simply vibrates and carvs the wave into a prepared soft disk. This disk is then hardened and used to create the metal print shape.

Id have to send them a mail to figure out what different samplerates would do to sound quality and in regard to pitching down the vinyl. Ill respond here with the answare, but it will not happen anytime soon, as I want to ask them a bunch of things at the same time. And im not in a hurry. Maybe someone else know?

[ChiefNugget]

Ive done some research about how they make vinyl. They make the print master disk with a needle. This needle draws the wave from the master file. Resolution and sample rates above 24/44.1might not be such a difference, since the wave is transfered with an "analog" mechanical arm to this needle. And the needle simply vibrates and carvs the wave into a prepared soft disk. This disk is then hardened and used to create the metal print shape.

Id have to send them a mail to figure out what different samplerates would do to sound quality and in regard to pitching down the vinyl. Ill respond here with the answare, but it will not happen anytime soon, as I want to ask them a bunch of things at the same time. And im not in a hurry. Maybe someone else know?

Cool. I'm in know hurry to find out either. It was just an interesting question. Thanks for taking the time to reply.

[ChiefNugget]

thats kind of like the .jpeg thing... where you don't increase a file size without artifacts, but you can decrease it and have a perfect, though smaller, copy.

Maybe Ableton can hire the CSI Miami guys to overcome this shortcoming of physics!

yeah, then they could hire them to tell us how to make a sandwich bigger after we've had a few bites.

or we could just accept the fact that diminishing file depths, and quality typically leaves better results than trying to increase and insert those settings later. YRMV.

They'd be like the Jesus of digital audio, creating samples from nil. If you've ever seen any of the (awful) CSI shows, you'll understand that those guys can pull it off

[3dot...]

what about 44 vs. 48 ???

[kevwestbeats]

I think its totally a preference thing but I personally do not go beyond 44.1 because I cannot and many pros cannot hear the difference and the ones that can say its not even a large difference.

[ze2be]

I think its totally a preference thing but I personally do not go beyond 44.1 because I cannot and many pros cannot hear the difference and the ones that can say its not even a large difference.

Simply explained, using 44.1 kHz you divide it from stereo to left and right, and ends up with a 22.000 Hz range. The human ear can hear from about 20 to 20.000 Hz, so it is a bit above that. As far as I understand, going higher just adds to the higher frequencies that we can not hear anyway. Like 48 kHz would be reaching up to 24.000 hz. Thats so high theres no information up there for the human ear. Maybe ants can hear it, I dont know! But for us, it might be a difference when the file is played at lover pitch, like DJs do. 44.1 is just about the limit of what we can hear. But speed that down a bit and the sample rate goes below this audible range. Thus it could make a difference on digital files, if not as much on vinyl.

Ive herd complaines about muddy kicks and lowered sound quality when pitching down 44.1khz CDs, when DJing.

Its easy to test this. With a good eq, try to isolate a range from 20khz to 22khz, (the highest Hz of 44.1khz) and see if you can hear a difference while muting it on and off in a full mix. Then try a range from 22khz to say 24khz, to simulate the difference between 44.1 and 48khz.

[doom_Oo7]

You all seem to forget, people, that one of the main benefitis of high samplerates, is that it prevents aliasing due to poorly designed virtual instruments.

(aliasing is when frequency is generated over the nyquist frequency and the sampling makes a "come-back" of these frequencies in reverse order. E.G a sound at 24kkhz when sampling for 20khz would make some noise @ 16khz)

However, it also puts more strain on the CPU.
However however, for people like me who play live with VSTs it can help to reduce latency a further bit.

And I have heard (but never found, so this is to take with much precaution) that some high-frequency content could have some psychoacoustic effect upon our brain but well!

[Electric Blue Studios]

if your computer can handle it, go for 88.2.

[Angstrom]

I think its totally a preference thing but I personally do not go beyond 44.1 because I cannot and many pros cannot hear the difference and the ones that can say its not even a large difference.

Simply explained, using 44.1 kHz you divide it from stereo to left and right, and ends up with a 22.000 Hz range. The human ear can hear from about 20 to 20.000 Hz, so it is a bit above that.

That is not a correct explanation. I'm not sure where you got that from. I think you misunderstood something you read online.

Briefly correcting : Both stereo channels are sampled 44.1, there is no "divide it stereo left and right to end up with 22khz".

The stereo sample is recorded and played back at 44.1 (both channels) because the sampling frequency needs to be double the highest audible frequency in order to prevent aliasing. All the aliases will occur above the Nyquist frequency.

Why : Imagine trying to sample a 22khz sine wave at only a 22khz sample rate. That is - one sample being taken in each sine wavelength . you would only get a sample of up, up, up, up, ( no down) . So you need double the highest audible frequency to represent that frequency sine wave, to get up down up down , you need to sample at least two points of the wave cycle. Double the frequency of the signal.

This has been the worlds simplest, and worst, explanation of sampling and Nyquist theorem. Thank you.

[ze2be]

This has been the worlds simplest, and worst, explanation of sampling and Nyquist theorem. Thank you.

Thanks, I stand corrected! Dont know much about it and dont claim I do either. Likely got it from here, and bits and pieces arround.

I just tried to get my head arround if higher sample rate would matter when the file is pitched down by a DJ, or a samplist, etc. Do you have any thoughts about that?

[doom_Oo7]

It depends, if for instance it uses a FFT it doesn't really matter since you won't work in the time domain but the frequency domain.
And well.. Just try ^^ pitch down heavily a file @ 44 khz and the same file @ 96khz

[H20nly]

However, it also puts more strain on the CPU.
However however, for people like me who play live with VSTs it can help to reduce latency a further bit.

if this is true, what are the chances that this bit of information could help the folks with PDC issues?

is it possible they are using a lower sample rate and that increasing it [significantly] would help with that?

[Dragonbreath]

I would suggest that you stick at 44,1 since all you samples are at that sample. Upconverting you will loose sound quality witch is counter productive. You will have to downsample back to 44,1 to distribute for playback anyways. The more times you change the sample rate the more the sound degrades.

44,1 was selected for CD production because thats all thats needed for full reproduction of audible frequencies. They say humans can hear 20hz to 20khz. Thats more then genereous range. The average adults hears even less then that. Try with a tone generator and good monitors or headphone see how far you can get. You need quite loud volumes to hear the extreme ends of the frequencies. Your lucky if you can hear bellow 30hz and over 16-18khz.

The main advantage would lower latencies. From my readings the higher sample rates are mostly marketing hype from soundcard manufactures.

Most playback systems are not designed to reproduce the frequencies that higher sample rate permit and can actually cause something know as intermodulation frequency, were the system creates alias in the audible frequencies trying to reproduce frequencies it cannot output.


Bottow line is I dont see any reason to use a higher sample rate expect for latency

[ChiefNugget]

However, it also puts more strain on the CPU.
However however, for people like me who play live with VSTs it can help to reduce latency a further bit.

if this is true, what are the chances that this bit of information could help the folks with PDC issues?

is it possible they are using a lower sample rate and that increasing it [significantly] would help with that?

I was wondering about the VSTi latency vs sample rate topic also. I thought that many (most?) good softsynths use oversampling to avoid aliasing. If they're already oversampling internally, would the output sample rate affect the latency?

[Tone Deft]

This has been the worlds simplest, and worst, explanation of sampling and Nyquist theorem. Thank you.

Thanks, I stand corrected! Dont know much about it and dont claim I do either. Likely got it from here, and bits and pieces arround.

I just tried to get my head arround if higher sample rate would matter when the file is pitched down by a DJ, or a samplist, etc. Do you have any thoughts about that?

it wouldn't matter. at the point in the creation process when the tone arm is scraping the audio into the master disc the signal is analog and any notion that it was ever digital is lost.

from what I've heard the guys at the Mastering labs are seriously anal and want the artist to stay completely away from the process. they ask for the track to be 3-6dB down from clipping and for the artist to stay away from the entire process. these guys are serious audio geeks with 10s or even 100s of thousands of dollars of gear that they know and love and the general population doesn't have a freaking clue how to use.

all the aspects of higher sampling rates have been covered in this thread. 48kHz 24 bit is fine. as I wrote before, the weakest link is SRC.