[skeptic]

My ears seem to confirm the majority opinion in this thread on audio asylum: Audio Asylum Thread Printer

You can experiment with changing the ASIO buffer value on the fly in WinAmp (while a track is currently playing), and this is no subtle change in soundstage! I was curious if anyone could explain the phenomena to me, how/why the buffer value impacts the width of the soundstage, apparently diminishing the audible width as the buffer value is increased?

Apparently, in some high end DAC's, like this Chord, the opposite is true? See: Chord Electronics DAC64 hi-fi+ technical review

Looking forward to your comments!

 

[Rempert]

I like how in the one link a poster claims that a professional musician won't notice an 80 millisecond delay, and in the other link the reviewer claims that human ears are sensitive to a delay of a few microseconds! I know from firsthand experience that the first claim is bunk, and the second probably defies some laws of physics.

That Chord DAC apparently has settings involving a filtering DSP, and to get the best filtering mode requires high latency. That's believable enough. Changing the latency settings in WinAmp or Foobar2000, however, does not change the DSP chain. If there is an audible difference in sound quality from changing these buffer settings, I would find that rather disconcerting. I've never really messed with those settings much except on my recording software, where I want low latency for the obvious reasons.

 

[skeptic]

Just thought I'd give this thread a bump to see if any other Winamp users have had an opportunity to experiment with the ASIO buffer value?

I notice a clear change in soundstage, expanding/contracting inversely with the ASIO buffer value, through my X-fi --> Woo --> Beyer 880 setup.

But maybe this is gear dependent? I wonder if the same change occurs through a USB DAC or when merely using the sound card as a digital transport to an external spdif DAC?

 

[skeptic]

Having recently added a KECES 131 to my setup, I figured I would add an update to my comments regarding Winamp ASIO buffer values and soundstage.

I had theorized (in my bubble of ignorance) that a better DAC might be less susceptible to the sonic impacts of increasing the buffer value.

However, to the contrary, I would say that the difference in soundstage caused by increasing/decreasing the ASIO buffer in Winamp is arguably even more pronounced through the KECES than it was through the X-fi's analog out.

The configuration options allow the buffer to be set anywhere from 0 - 63.

To my ears, good soundstage is produced in the 5-8 range, but if you even push it up to 12, the sound becomes seriously compacted.

I would love to understand why this happens. Where's a Headphone Supremus to break it all down in layman's terms when you need one?

 

[b0dhi]

Quote:

Originally Posted by Rempert /img/forum/go_quote.gif I like how in the one link a poster claims that a professional musician won't notice an 80 millisecond delay, and in the other link the reviewer claims that human ears are sensitive to a delay of a few microseconds! I know from firsthand experience that the first claim is bunk, and the second probably defies some laws of physics.

The first is clearly nonsense, but the second is actually true. The ear can tell an interaural time difference of between 2-15 microseconds (depending on the ear and on the input).

I've never noticed a difference in anything by changing the buffer size. Except ofcourse when there's stutter when the buffer is too small. I'd be very very surprised if raising the buffer actually dimished the sound quality in any way whatsoever.

 

[skeptic]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif The first is clearly nonsense, but the second is actually true. The ear can tell an interaural time difference of between 2-15 microseconds (depending on the ear and on the input). I've never noticed a difference in anything by changing the buffer size. Except ofcourse when there's stutter when the buffer is too small. I'd be very very surprised if raising the buffer actually dimished the sound quality in any way whatsoever.

If your definition of "sound quality" includes the width of the sound stage, I promise you that it does. I don't understand why, but hearing is believing...

Winamp and ASIO are freeware applications/drivers. Maybe you could DL them both, give it a try, and let me know what you hear? You can change the buffer value on the fly, mid-song, so A/Bing it is really really easy.

Maybe its something about the Woo/Beyer combo...but both the X-fi DAC and my KECES reveal the same relationship between buffer value and soundstage.
Honestly, it's "very very supris[ing]" to me as well

 

[eraser_svk]

And what is the best value ? Are 100 ms ok ?

Does this also occur on a foobar 0.8.3 ?

 

[b0dhi]

Quote:

Originally Posted by skeptic /img/forum/go_quote.gif If your definition of "sound quality" includes the width of the sound stage, I promise you that it does. I don't understand why, but hearing is believing... Winamp and ASIO are freeware applications/drivers. Maybe you could DL them both, give it a try, and let me know what you hear? You can change the buffer value on the fly, mid-song, so A/Bing it is really really easy. Maybe its something about the Woo/Beyer combo...but both the X-fi DAC and my KECES reveal the same relationship between buffer value and soundstage. Honestly, it's "very very supris[ing]" to me as well

Hm I should have mentioned that I have been using Winamp/ASIO for years, and have never noticed any change in sound when increasing the buffer size beyond what's needed for stutter-free playback.

 

[regal]

My WINAMP+ASIO sends a bitperfect SPDIF which is decoded by my HDCD DAC. If there were any change in soundstage then the HDCD light would not turn on.

My guess is you who hear this are not bitperfect, sometimes the ASIO stream is messed up by the hardware or OS. IOW's not all ASIO is truly bit-perfect.

Take the HDCD or DTS test and get back to us with some real data.

 

[d-cee]

I don't hear a difference

Winamp > ASIO > E-MU 0404 PCI

OT: Quote:

Originally Posted by regal /img/forum/go_quote.gif My WINAMP+ASIO sends a bitperfect SPDIF which is decoded by my HDCD DAC. If there were any change in soundstage then the HDCD light would not turn on.

Are you saying your DAC knows the difference between bit perfect and not?

 

[regal]

Quote:

Originally Posted by d-cee /img/forum/go_quote.gif I don't hear a difference Winamp > ASIO > E-MU 0404 PCI OT: Are you saying your DAC knows the difference between bit perfect and not?

Yes absolutely, HDCD will not signal without bitperfectness. This is how the PMD100 HDCD filter works. If you don't understand this you need to brush up on digital theory, sorry D-Cee

 

[skeptic]

Quote:

Originally Posted by regal /img/forum/go_quote.gif My WINAMP+ASIO sends a bitperfect SPDIF which is decoded by my HDCD DAC. If there were any change in soundstage then the HDCD light would not turn on. My guess is you who hear this are not bitperfect, sometimes the ASIO stream is messed up by the hardware or OS. IOW's not all ASIO is truly bit-perfect. Take the HDCD or DTS test and get back to us with some real data.

Can you further explain the HDCD and/or DTS tests? How do I do this?

I'd love to understand what I'm hearing. Also, just fyi, the change is easier for me to observe on classical/folk - the less produced the better.

Maybe this is just some jitter induced artifact that has given my ears the illusion of wider sound stage due to buffer related timing issues manifested through my PCI bus or X-fi.

But I can confirm that my software settings are all correct for a bit perfect output... I'm even using a decent cable with impedance matched connector to avoid having the bits bounce on route to my DAC.
The KMixer is clearly being bypassed, and I'm in music creation mode, with bit-matched selected, so there's no alteration in sampling. I will also note that I get the same effect both through the Creative ASIO driver (as the X-fi has native ASIO support) and through ASIO4ALL. [The two are distinct options in the winamp output tab.]

If you can tell me how to run the tests you mentioned above without investing in any additional hardware, I'm game to try.

And thanks to both of you for the continued comments!

 

[Covenant]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Yes absolutely, HDCD will not signal without bitperfectness. This is how the PMD100 HDCD filter works. If you don't understand this you need to brush up on digital theory, sorry D-Cee

Actually, the act of believing that the DAC will recieve a bit-perfect signal regardless of settings forms a gateway whereby the consious self communicates with the younger self to create the desired outcome as potential energy in the quantum sea. An affirmation - such as posting on a forum that the DAC will definitely recieve a bit-perfect signal irrespective of buffer size - allows the potential energy formed in the quantum sea to manifest on the prime material.

If you don't understand this you need to brush up on your consensus reality theory, sorry regal

 

[upstateguy]

Hi Guys, Hi Skeptic

I tried this too, and didn't hear anything worth reporting by changing the ASIO buffer size, but I did hear an audible difference when I changed the Kernel Buffers back to 2 from 4... the most noticeable change was an increase in the treble response... it seemed to get a little louder and more etched.... does anybody else hear this?

my current configuration is

and my set up is: dedicated DIY computer>foobar 9.5(SRC at 44100)>ASIO 2.8>Stello DA100(upsampling off)>GS-1>701s (at the moment)

USG

d-_-b

 

[d-cee]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Yes absolutely, HDCD will not signal without bitperfectness. This is how the PMD100 HDCD filter works. If you don't understand this you need to brush up on digital theory, sorry D-Cee

if you're changing the SPDIF signal to HDCD format then it's automatically not bit perfect

if you're trying to claim that a DAC knows whether it's being fed a bit perfect FLAC file as opposed to a 128kbps lossy mp3 and stops working on the latter I think you're the one that needs brushing up on digital theory