Head-Fi.org › Forums › Equipment Forums › Sound Science › Blind test: 6 DACs compared
New Posts  All Forums:Forum Nav:

Blind test: 6 DACs compared - Page 8  

post #106 of 176
Quote:

Originally Posted by Hi Rez View Post

 

Actually I am not worried about audible differences that shouldn't be there.  In my first post I stated that I didn't hear any....

Don't tell me you're worried about inconsistencies causing files to sound similar...

 

I really don't think volume matching is going to have an effect aside from volume difference.

post #107 of 176
Quote:
Originally Posted by skamp View Post

I think you should worry about the influence of solar flares on my electronic gear, as well.

 

Just making the point that the value of any conclusions drawn from tests are only as good as the tests themselves - which I believe I have seen mentioned numerous times elsewhere in Sound Science.  And also that any tools used in tests need to be validated that they have no influence on the test.  

 

Thank you for creating this interesting exercise, however until you can show that your tools don't influence the results, we probably can't draw anything conclusive from it.  

 

Just curious though, how do you expect to have meaningful discourse in a thread with replies like that?

post #108 of 176
Quote:
Originally Posted by chewy4 View Post

I really don't think volume matching is going to have an effect aside from volume difference.

Remember that this is Sound Science. "Think" isn't good enough here.  

post #109 of 176
Quote:
Originally Posted by Hi Rez View Post

Remember that this is Sound Science. "Think" isn't good enough here.

True, but given that it's a function in a reliable and well established program program that isn't designed to cause any other changes then it's safe to assume that it doesn't. Unless someone has proof that it does.

post #110 of 176
Quote:
Originally Posted by nick_charles View Post

Quote:
Originally Posted by Hi Rez View Post

With any digital editor, even adjusting the volume levels requires bit manipulation and dithering.

 

But the original sample was edited with Audacity....

 

 

In another thread in this subforum we saw some evidence that even converting a 24 bit file to 16 bits with the worst possible type of dither and truncation did not do any audibly detectable damage...

Woah really? Could you post or PM the link? I've been kind of interested if I can hear the difference between so-called HD music and "regular" 'ol CD-quality music.

Sorry to derail the thread.

 

I haven't done any blind tests yet, but if there are any differences between the samples, I can't really distinguish them. I think I might be able to hear some subtle differences, but until I do an ABX test, I say they're all sonically identical to me.

post #111 of 176
Quote:
Originally Posted by Hi Rez View Post

With any digital editor, even adjusting the volume levels requires bit manipulation and dithering.

 

The original sample was not modified in any other way than being cropped to a length of 30 seconds. This does not require any "bit manipulation and dithering", and it can also easily be verified that the operation is lossless (other than obviously making the sample shorter).

 

Volume matching the recorded samples at 16-bit resolution does involve dithering, but this does not have a realistic chance of making the sound more similar to the original through some magic "synergy" with the DAC and ADC. In fact it just adds more noise and therefore increases the difference, but apparently not to an extent to make it audible.

post #112 of 176
Quote:
Originally Posted by stv014 View Post

The original sample was not modified in any other way than being cropped to a length of 30 seconds. This does not require any "bit manipulation and dithering", and it can also easily be verified that the operation is lossless (other than obviously making the sample shorter).

 

I believe it would depend on how the Audacity preferences are set whether dithering would occur.  http://forum.audacityteam.org/viewtopic.php?f=12&t=9429

post #113 of 176

Right, I do not normally use Audacity, but with the default settings, version 2.0.0 does indeed seem to convert the input file to floats first, and then the 16-bit output file is dithered and not bit perfect. Not only that, but it even uses a poor and broken "noise shaped" dither that makes the noise more audible, in addition to increasing the overall level (it is because when dithering multichannel files, the same state information is shared by all channels, and this breaks the algorithm; it is an old bug, and apparently still not fixed in the version I tested). It has an A-weighted dynamic range of only ~81.5 dB, or about 13.5 bits.

 

Edit: setting the default sample format in "Preferences: Quality" to 16-bit and/or disabling dithering fixes the problem. I do not know what version and settings skamp used, but if the "original" sample was unintentionally made noisy, it can be found out easily. Having said that, I doubt that fixing any Audacity related problems would suddenly reveal major differences between the DACs.


Edited by stv014 - 2/1/13 at 3:03pm
post #114 of 176

No dithering is required without level, EQ or bitrate modification. When dither is used on a level change it's there to keep the file sounding the same, not make it different unless it's not right. I would disagree with those that don't notice different dithers under the best circumstances.

 

 

I think Skamp did this so that we should listen and then examine and compare results. There's all kinds of outs here if you want them but that's hardly the point. Lets see if there's any corroboration of results first and then we can discuss why or why not. 

 

I would personally not use a USB a2d as a means for critical compare but it's fine for this and it's not my test. I think it's cool he's gone to the trouble. We don't know about the original rip or quality of master either. So what? Skamp never claimed it was super critical test and some differences should still be there. The files are fine enough for this. Take the test and we can discuss why the results turned out as they did when they're in. Have some fun and see what you get.


Edited by goodvibes - 2/1/13 at 9:09pm
post #115 of 176
Quote:
Originally Posted by miceblue View Post

Woah really? Could you post or PM the link? I've been kind of interested if I can hear the difference between so-called HD music and "regular" 'ol CD-quality music.

Sorry to derail the thread.

 

I haven't done any blind tests yet, but if there are any differences between the samples, I can't really distinguish them. I think I might be able to hear some subtle differences, but until I do an ABX test, I say they're all sonically identical to me.

 

 

http://www.head-fi.org/t/643951/can-you-hear-the-difference

post #116 of 176
Thread Starter 
Quote:
Originally Posted by Hi Rez View Post

Just curious though, how do you expect to have meaningful discourse in a thread with replies like that?

I reserve the right to be sarcastic as I see fit.

I cut and saved the original sample in Audacity as 24 bit. Then I computed a replaygain value of it, which I applied to the file while converting it to 16 bit at the same time, with SoX, with dithering and noise shaping enabled.

I made all recordings at 24 bit, as the EMU doesn't do 16 bit. Then I sample aligned them in Audacity and saved them as 24 bit. Then I applied the same treatment as with the original sample (replaygain and SoX).

Again, you're welcome to provide ABX logs demonstrating that such treatment either makes the end result audibly different, or makes original differences disappear (aside from volume, obviously).
Edited by skamp - 2/2/13 at 2:10am
post #117 of 176

Re. how valid this test is, because it relies on the playback facilities people have:

 

- It's possible that inferior device A might have faults that cancel out the particular ones of device B. If it is the case for a particular combination then it should easily show up, especially if each listener tests with a couple or more DAPs.

 

- DAP quality may act as a bottleneck more than a multiplier, ie if have devices A > B > C > D, where A is best, then when they are played back on C they'll all be reduced to a C level of quality, except D, which will still relatively suck. But again if that happens, it should be easy to spot from the pattern of results ***and in itself it will tell use a lot.***

 

So while I understand the objections some people made, I do think that they don't stop this from being a valid experiment.

 

For myself, I just listened using the built-in sound of my Shuttle XS35 book-sized PC (which is probably about as good as my Clip or J3) using the old pair of HD212Pro's I keep connected - I don't think I could really tell the difference between any of the tracks, although 6 of them are overwhelming, at least for me. When I know which track is which I'll get someone to blind test me with, say, the Clip and 02/ODAC. And I'll repeat with my J3 and HD25s and my own Clip and HF5s.


Edited by scuttle - 2/3/13 at 8:00pm
post #118 of 176

Suggestion to Skamp: Have you considered picking say two players A and B and repeating the play record cycle several times on each? I.e. take your Clip, play & record - and use the result as the input to the next cycle WITH THE SAME PLAYER, then repeat several times? This could be valuable in two ways:

 

 

1. It would tell us whether quality tends more to be a bottleneck effect or multiplicative (if you can't hear SQ dropping off it's more of a bottleneck effect.)

 

2. If quality is multiplicative, then this procedure should be very revealing of the differences:

 

...0.95 accuracy raised to the power of 6 is 0.74

 

...0.9 ^ 6 is 0.53

 

So a 1 in 20 difference has become a 1 in 2 difference!

post #119 of 176
Thread Starter 
scuttle: thanks for the suggestion, but I really can't be bothered to do that tongue.gif It would be somewhat interesting, but it wouldn't help with the purpose of this test.
post #120 of 176

Was not an easy test at all. All the samples sound very close upon initial listening.

After listening for good 25-30 min here is my opinion (Hope C is not some portable playereek.gif ) - 

 

A - Thin but good

B - congested
C - very good
D - good
E - congested
F - congested
G - ok
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
This thread is locked  
Head-Fi.org › Forums › Equipment Forums › Sound Science › Blind test: 6 DACs compared