Scoring Central
Develop samples at 48 khz Samplefrequency - Printable Version

+- Scoring Central (http://scoringcentral.mattiaswestlund.net)
+-- Forum: Technology (http://scoringcentral.mattiaswestlund.net/forumdisplay.php?fid=5)
+--- Forum: Recording & Sampling (http://scoringcentral.mattiaswestlund.net/forumdisplay.php?fid=9)
+--- Thread: Develop samples at 48 khz Samplefrequency (/showthread.php?tid=383)



Develop samples at 48 khz Samplefrequency - l33 - 04-22-2018

Hi,
please develop exclusively samples at 48 KHZ samplefrequency.

48 KHZ is perfect and if some one would really need 44.1 khz, this can be downsampled, I mean converted to 44.1 KHZ.

But if you work at 48 khz, and you use 44.1 khz samples, you will maybe not notice directly, but constant real time sample rate conversions are going on.
These are maybe not audible first, but when listening closer, I think, some kind of hissing and phasing can be heardt, thats why I am asking to think about recording and releasing samples at 48 khz.
44.1 khz is cd quality and surely great sounding, but TV and dvds are at 48 khz I think, and I really think it does not hurt recording at 48 khz.


RE: Develop samples at 48 khz Samplefrequency - Samulis - 04-22-2018

The upscaling of 44.1 kHz audio to 48 kHz should have no audible effects, unless you are using a super outdated or poorly scripted audio engine within the DAW. It is alternately likely that what you are hearing is most likely confirmation bias, that is, your own mind tricking you into thinking there is a difference. It is a very common issue- the only way to know for sure is to blind listen to the tracks (i.e. obscure which is which) and see what result comes out. You would need to be able to tell which is which without looking at the files in any way 10/10 times.

You can also do a common trick, by inverting one file and layering it over the other, exactly in sync, which will reveal the "difference" between the files. If you can hear that difference at a typical listening level, then there is probably something wrong with your software's converter.

You can see some comparisons of various DAW's converting from 96 kHz to 44.1 kHz for example here-
http://src.infinitewave.ca/

Pick two DAWs and it will show comparisons of their conversion filters and accuracy. You can see some do great jobs (e.g. Adobe Audition) while others do less so. Regardless, almost all of these would not have audible consequences unless you were filtering out the audio then amplifying the background noise by 60 dB or so- or otherwise listening loud enough to blow your eardrums out.

Besides, if there is a problem with upscaling 44.1 to 48, then there is also going to be a problem downscaling from 48 to 44.1 for anyone who uses 44.1! Wink

There is no real argument that 48 kHz or 44.1 kHz is better. They are both more than suitable to cover the entire and complete range of human hearing, with much to spare (most people around 20+ years old can only hear up to around 18 kHz, especially if, like most musicians, they are reckless and listen too loudly to music). The only difference is in end use- one is standardized to work on CD audio, the other on DVD/video audio. As the general users of virtual instruments is a mix of film composers and artists/musicians, it is impossible to make a convincing argument for either. Typically, many companies do the original sampling at a higher sample rate, 'just to be safe', then downscale for distribution, but this not always true.

For the developer, one could even make an argument that 44.1 kHz can save bandwidth. The difference is about a 9% saving, which over the course of say, 40 GB, might mean just 36 GB being needed instead. This is not major, but every bit helps in an economy of scale situation, such as distributing several TB of data every month, as most developers do. At ~9 cents per GB, distributing 10 TB of data at 44.1 instead of 48 would mean about $82 in savings. It is not huge, but it helps, especially when developing freeware, where unnecessary costs should be avoided as much as is possible.


RE: Develop samples at 48 khz Samplefrequency - Mattias Westlund - 04-23-2018

Unless you have superhuman hearing or, like Samulis says, there is something wrong with the way the audio is being converted or reproduced, there is no way that anyone can hear a difference between 44.1 and 48 kHz. Sure, some instruments will benefit from a higher sampling frequency (especially stuff with a lot of high end, like cymbals or steel-string acoustics), but in that case you will need to go to 88.2kHz or higher for there to be any audible difference. And not even at 192kHz is the difference so dramatic that it's worth the hassle of dealing with a much increased file size and overall system load. I know that some people will use 96 or 192 kHz 32-bit FP recordings for archival purposes, but that's about it.

And while we're at it -- no, you can't hear a difference between 16-, 24- or 32-bit recordings either, not at optimal levels at any rate.


RE: Develop samples at 48 khz Samplefrequency - l33 - 04-24-2018

Thanks for the answers. I made a test online some time ago and I think I can hear up to 17khz but not higher, so , yes, I surely can not hear a difference.
I do not know about psychoacoustics and waveform interpolation - but we all know there are people saying analog sounds better...
Yes, sometimes, when hearing crash cymbals, I think I can hear that its digital...specially when listening on headphones.
Again, I never said 48 khz sound better, though yes, sometimes I think so...but...I think sample rate conversion is reducing realtime performance and it surely never makes the sound better...


RE: Develop samples at 48 khz Samplefrequency - Mattias Westlund - 04-24-2018

(04-24-2018, 06:39 PM)l33 Wrote: Yes, sometimes, when hearing crash cymbals, I think I can hear that its digital...specially when listening on headphones.

I think that is probably the result of lossy audio compression rather than sampling frequency, unless of course you listen to all your music in lossless formats. But like I said, cymbals and other trebly instruments do sound smoother at higher sampling frequencies... but you need to go way higher than 48 to get a noticeable difference. And even then, it's slight.

(04-24-2018, 06:39 PM)l33 Wrote: Again, I never said 48 khz sound better, though yes, sometimes I think so...but...I think sample rate conversion is reducing realtime performance and it surely never makes the sound better...

Yes, but the vast majority of DAW systems will be running at 44.1kHz so what does moving to 48kHz solve? Doesn't it make more sense to simply stick with 44.1 since it is pretty much the standard outside DVD's (which are going the way of the dodo anyway )?