[FlyingKiwi]

Hi
 
I'm trying to do some ABX testing using Foobar 2000, and also to create a simple set of steps for other people to do the same.
 
Although in my case, I output from Foobar to an external DAC which displays the incoming sample frequency rate and bit depth, I know that many people don't have external DACs and want to be able to route the Foobar output through their computer's sound card and then probably to headphones.
 
So my question is, if we assume we might be testing two pieces, one at 44.1/16 and the other at 96/24, how can we be sure that the output from Foobar that is fed into the soundcard is not changed either by Foobar or Windows prior to entering into the sound card's DAC?
 
I've loaded the various components for KS and WASAPI.  But - at least for my computer - it seems nothing in it supports Kernel Streaming, and if I choose WASAPI, there's a requirement to then specify the output format bit depth, and my fear is that this mandatory setting will override the bit depth of the music pieces.
 
Clearly, ensuring that the test pieces make it all the way to the sound card's DAC unchanged is an important part of ensuring valid ABX test results.
 
If anyone can advise how to be sure this is occurring correctly, it would be very helpful and much appreciated.

 

[stv014]

Originally Posted by FlyingKiwi /img/forum/go_quote.gif
 
So my question is, if we assume we might be testing two pieces, one at 44.1/16 and the other at 96/24, how can we be sure that the output from Foobar that is fed into the soundcard is not changed either by Foobar or Windows prior to entering into the sound card's DAC ?

 
It is actually recommended to convert the 44.1/16 sample in software to 96/24 first, and then compare files of identical format. Otherwise, there could potentially be a difference due to the playback hardware (even if it should ideally be transparent, it is best to limit the number of variables in the test), and also spurious clues, such as the DAC outputting a click when the sample rate is switched - I can confirm this happening with my sound cards.
 

Originally Posted by FlyingKiwi /img/forum/go_quote.gif
 
I've loaded the various components for KS and WASAPI.  But - at least for my computer - it seems nothing in it supports Kernel Streaming, and if I choose WASAPI, there's a requirement to then specify the output format bit depth, and my fear is that this mandatory setting will override the bit depth of the music pieces.

 
Kernel streaming is used in Windows 2000 and XP, while WASAPI is in Vista and newer, so the former is outdated. Use WASAPI at 96/24, and upsample the lower resolution sample with a good converter.

 

[castleofargh]

I have the lazy worst case scenario that could be used alternatively with stv014's suggestion to test different thing. 
just set 24bit output on wasapi and that's it. the 16bit will receive it's fair share of added zeros in every sample, but that won't change the sound in anyway from zero to -96db the signal will be exactly the same 16bit signal. (minus the quantization noise the dac should add. but that's only in case your dac doesn't already add the zeros itself without telling you anyway).
 
that way you'll be testing the files pretty much as is, and also how your dac handles different sample rates, so it might not be what you wish to test all the time, but for product testing it can in fact be a nice perk to see if the sample rate brings audible differences on one dac or another.

 

[ab initio]

It is actually recommended to convert the 44.1/16 sample in software to 96/24 first, and then compare files of identical format. Otherwise, there could potentially be a difference due to the playback hardware (even if it should ideally be transparent, it is best to limit the number of variables in the test), and also spurious clues, such as the DAC outputting a click when the sample rate is switched - I can confirm this happening with my sound cards.


Kernel streaming is used in Windows 2000 and XP, while WASAPI is in Vista and newer, so the former is outdated. Use WASAPI at 96/24, and upsample the lower resolution sample with a good converter.

I have the lazy worst case scenario that could be used alternatively with stv014's suggestion to test different thing. 
just set 24bit output on wasapi and that's it. the 16bit will receive it's fair share of added zeros in every sample, but that won't change the sound in anyway from zero to -96db the signal will be exactly the same 16bit signal. (minus the quantization noise the dac should add. but that's only in case your dac doesn't already add the zeros itself without telling you anyway).

that way you'll be testing the files pretty much as is, and also how your dac handles different sample rates, so it might not be what you wish to test all the time, but for product testing it can in fact be a nice perk to see if the sample rate brings audible differences on one dac or another.

These are both fine suggestions.

Cheers