Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: 192khz samples? (Read 5937 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

192khz samples?

There are a few "audiophile" sample sites out there that record their instrument samples at 192khz. They sell the raw version at 192khz, and downsampled version at 44.1khz. They say they used Wavelab6's crystal resampler to downsample. Their 192khz editions pretty much have to be resampled by the end user for every project.

My question is, are commerical resamplers today (expensive ones like Wavelab 6), pretty much all transparent today?

http://src.infinitewave.ca/

Are these comparisons of theoretical value only?

192khz samples?

Reply #1
All resampling isn’t transparent. Some of the hardware resampling done on earlier gaming type soundcards produced enough noise and distortion to be noticeable under some conditions, but it might be that the large measured variances in different software products displayable at that link are mostly irrelevant.

Software resamplers do not need to be expensive to be good. The inexpensive version of CoolEdit was doing exceptionally clean resampling more than ten years ago. There are currently some freeware resamplers that are considered very good. I’ve never seen mention of any ABX tests for audible differences between the results of different software.

192khz samples?

Reply #2
There are a few "audiophile" sample sites out there that record their instrument samples at 192khz. They sell the raw version at 192khz, and downsampled version at 44.1khz. They say they used Wavelab6's crystal resampler to downsample. Their 192khz editions pretty much have to be resampled by the end user for every project.

My question is, are commerical resamplers today (expensive ones like Wavelab 6), pretty much all transparent today?

http://src.infinitewave.ca/

Are these comparisons of theoretical value only?


I suspect that the Sound Forge resampler without any anti-aliasing at all might fail an ABX.

There's a ton of resamplers at the Infinite Wave web site, and I haven't been through them all.

I suspect that mishandling of aliasing and the transition band could leave audible artifacts. But, the transition band problems would have to be pretty sever. No antialiasing at all is a pretty severe problem, IMO.

192khz samples?

Reply #3
There are a few "audiophile" sample sites out there that record their instrument samples at 192khz. They sell the raw version at 192khz, and downsampled version at 44.1khz. They say they used Wavelab6's crystal resampler to downsample. Their 192khz editions pretty much have to be resampled by the end user for every project.

My question is, are commerical resamplers today (expensive ones like Wavelab 6), pretty much all transparent today?

http://src.infinitewave.ca/

Are these comparisons of theoretical value only?


I suspect that the Sound Forge resampler without any anti-aliasing at all might fail an ABX.

There's a ton of resamplers at the Infinite Wave web site, and I haven't been through them all.

I suspect that mishandling of aliasing and the transition band could leave audible artifacts. But, the transition band problems would have to be pretty sever. No antialiasing at all is a pretty severe problem, IMO.

So downsampling isn't a simple matter of loading the files into a software and clicking the button? Like load a 192/24 track, select 44.1/24 and click "start" or something like that?

192khz samples?

Reply #4
I'll just talk about CoolEdit. There not being anything demonstrably better, I've had no need to look at the others' interfaces.

Beside choosing the target sample rate, you check a box that tells it to use the pre & post filters. That should always be done unless you are running some kind of experiment, and it alwasy stays in the last state marked, so after the first time setup you can ignore it.

The other consideration is the filter "quality." This is continuously variable from 30 to 999. The wisdom from the developers forum is that 250 produces the highest possible quality resampling. Larger values just effect the unptheenth decimal place and add significantly to the calculation time, while giving you nothing useful in return. Your choice. This also stays at the last used setting until changed again.

So, in essence, once it is set up, the operation is: select that function (called Convert Sample Type in CookEdit), select the target sample rate, click on OK. Other programs are probably similar, i.e. not very complicated.

If you wanted to adjust the bit depth (i.e. convert to 16 bit to make an audio CD) there are a couple other options effecting only that part.

There is also an option to convert two channels to one or one channel to two (the two will be identical) but this is obviously unrelated to sample rate conversion.

192khz samples?

Reply #5
I'll just talk about CoolEdit. There not being anything demonstrably better, I've had no need to look at the others' interfaces.

Beside choosing the target sample rate, you check a box that tells it to use the pre & post filters. That should always be done unless you are running some kind of experiment, and it alwasy stays in the last state marked, so after the first time setup you can ignore it.

The other consideration is the filter "quality." This is continuously variable from 30 to 999. The wisdom from the developers forum is that 250 produces the highest possible quality resampling. Larger values just effect the unptheenth decimal place and add significantly to the calculation time, while giving you nothing useful in return. Your choice. This also stays at the last used setting until changed again.

So, in essence, once it is set up, the operation is: select that function (called Convert Sample Type in CookEdit), select the target sample rate, click on OK. Other programs are probably similar, i.e. not very complicated.

If you wanted to adjust the bit depth (i.e. convert to 16 bit to make an audio CD) there are a couple other options effecting only that part.

There is also an option to convert two channels to one or one channel to two (the two will be identical) but this is obviously unrelated to sample rate conversion.

According to that SRC comparison site, does the Crystal resampler or the Izotope 64 look better in those graphs?

192khz samples?

Reply #6
I checked two of the IZotope 64 sweep plots. They are good examples of how it should look.

The Gold Wave plot is a good, but not the most extreme, example of what you should not get.

Do they sound different? ABX testing is the way to tell.

I see nothing listed for a "Crystal resampler."

192khz samples?

Reply #7
I checked two of the IZotope 64 sweep plots. They are good examples of how it should look.

The Gold Wave plot is a good, but not the most extreme, example of what you should not get.

Do they sound different? ABX testing is the way to tell.

I see nothing listed for a "Crystal resampler."

What about the other graphs (other than sweep).

The crystal resampler is a part of wavelab 6, and is listed under wavelab 6.

192khz samples?

Reply #8
Compare them to the Adobe Audition 3 (with filters) results. If they look similar, they are good.

192khz samples?

Reply #9
I think major deviations seen in the sweep trace of some , aliasing and harmonic distortion, are the aspects most likely to be audible.

 

192khz samples?

Reply #10
The graphs may show that a software is working well and doesn´t produce unwanted garbage. What graphs look good lays in the eye of the beholder.
Some will say you must go close to Nyquist with high steepness others don´t. Some will say you need a bit alias and don´t need the full banwith to get a nice impulse. Others even want to change the absolute phase. Like often in audio you can´t have all and the recommendations differ.
Most likely all the resamplers on infinitewave sound the same as long they don´t cut HF to early or do some strong aliasing.
I didn´t find much conclusive things about that, leave alone conclusive abx tests!
I for myself decided to use sox cause it does a clean job and is free. I don´t keep anything above 20kHz and even allow aliasing up there. I found that this way i don´t have to worry about clipping also. If the original is maxed to 0dB and i use Shibatch´s SSRC for example i have to lower the volume up to 0.5dB to prevent clipping.

Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!