Audible Differences in DAC's? Options

[sawdin]

When he announced the release of the ODAC, NwAvGuy posted the the following on his blog:

NO SNAKE OIL REQUIRED: Many audiophiles want to believe more elaborate or exotic DACs offer higher fidelity. The ODAC demonstrates you do NOT need any of these for 100% transparent performance:

Asynchronous USB
UAC2 (USB Audio Class 2) Support
Asynchronous Sample Rate Conversion (ASRC),
Minimum Phase Filtering (no pre-ringing)
Non-oversampling NOS DAC chips
Dual DAC chips
Balanced Outputs
Vacuum Tube Stages
Elaborate and/or High Current Power Supplies

Is his position accurate/defensible?

If there was a well-designed DB/ABX test of the ODAC and sub $500/$1000 DAC's (e.g., Audioquest Dragonfly, iFi Micro iDAC, Audioengine D1, Emotiva XDA-2, etc.) that are are Asynchronous, USB Audio Class 2, etc. etc., do you think there would be 'audible/discernible' differences?

The reason that I am asking is that I am curious as to the whether there really is much of an audible difference between various sub $500 or $1,000 DAC's, and I haven't come across any DB tests of DAC's. I have read many subjective reviews that claim differences are quite noticeable. Of course, there are major differences in terms of the types and number of inputs and outputs that a DAC may have, whether they include a headphone amp, how portable they are, etc.

I am not trying to start a fight/flame war, I am just curious as to what the 'prevailing wisdom is' on this site. If you think there are differences in sub $500/$1000 DAC's, what do you believe are the relevant factors (e.g., USB vs Toslink/Coax)? I know that 'speakers' make a difference, but am unsure about how much, if any, difference there is between DAC's. Again, just trying to become an informed consumer (and save $), not trying to start fights.

TIA

PS...I should note that I appreciate people like NwAvGuy and others who try to provide quality products, diy options, etc., at reasonable prices.

[ktf]

You can try for yourself, download this: http://dl.dropbox.com/u/5985984/DAC-test.zip

It's a blind test I made with a Sound Blaster X-Fi Surround 5.1 USB-DAC which costed me around 50 euro. The first test has two files, one is the original, the other is that original after it has been played and recorded (looped back over a short unbalanced cable) 24x. So, any degradation coming from the DAC and ADC in that thing has been added and stacked 24 times. Try to find any differences.

The second test is another piece of music, with a few files looped back 4 times, 8 times and 12 times IIRC, and two on which I just applied some random (small) gain to make it a little harder to get the answer right if you're just guessing.

Even this cheap stuff performs very well. I don't think you'll hear any differences at all. Enjoy.

This post has been edited by ktf: Jan 9 2013, 22:57

[arve]

You can try for yourself, download this: http://dl.dropbox.com/u/5985984/DAC-test.zip

It's a blind test I made with a Sound Blaster X-Fi Surround 5.1 USB-DAC which costed me around 50 euro. The first test has two files, one is the original, the other is that original after it has been played and recorded (looped back over a short unbalanced cable) 24x. So, any degradation coming from the DAC and ADC in that thing has been added and stacked 24 times. Try to find any differences.

I haven't had time to look at the second set, but on test test1-1 vs test 1-2, it was really easy to discern the difference between the two - the noise floor in test1-2 was very different from the one in test1-1. I think it took me about a minute to do ten trials.

Test done using ABXer on OSX, with the FLAC files converted to .wav using ffmpeg in order to have playable files. Test log follows:

CODE

ABX Test Completed: 2013-02-04 19:28:17 +0000

Number of tests performed: 10
Number of correct answers: 10
Percentage correct: 100%

File 1 = /Users/arve/Downloads/DAC-test/wav/test1-1.wav
File 2 = /Users/arve/Downloads/DAC-test/wav/test1-2.wav
File placement was static.

n [A] [X] [B] Choice Score
1 [2] [2] [1] A 1/1
2 [2] [2] [1] A 2/2
3 [2] [2] [1] A 3/3
4 [2] [2] [1] A 4/4
5 [2] [1] [1] B 5/5
6 [2] [2] [1] A 6/6
7 [2] [1] [1] B 7/7
8 [2] [2] [1] A 8/8
9 [2] [1] [1] B 9/9
10 [2] [2] [1] A 10/10

--------------------------------------------------------------

[mzil]

At extremely elevated levels (which with normal level music signals would be deafeningly loud and clip any external amplifier), I would assume the noise floor of almost any two devices or circuits, which differs by one or more dB, would be discernible with musical silence as the test signal. That means nothing in real world use though.

[arve]

At extremely elevated levels (which with normal level music signals would be deafeningly loud and clip any external amplifier), I would assume the noise floor of almost any two devices or circuits, which differs by one or more dB, would be discernible with musical silence as the test signal. That means nothing in real world use though.

Note, I didn't perform this test with a high volume at all - listening volume was entirely typical for daily listening - average SPL when playing a pop recording is between 75-80 dB(A) in the listening position. Setup is near-field using a pair of M-Audio BX5 D2's, so nothing particularly high end either.

I also had a look at the first two files from the second test set, and based on those two files the test set is unsuitable for ABX testing because of level differences that are trivially detectable using rapid-switching ABX (I typically switch 1-3 times/second during a test, and alternate between A and X and B and X).