Head-Fi.org › Forums › Equipment Forums › Sound Science › BLIND TEST: onboard DAC
New Posts  All Forums:Forum Nav:

BLIND TEST: onboard DAC

Poll Results: Can you hear the difference ?

This is a multiple choice poll
  • 13% (2)
    I can hear the difference between A and B
  • 20% (3)
    I cannot hear the difference between A and B
  • 0% (0)
    I can hear the difference between A and C
  • 13% (2)
    I cannot hear the difference between A and C
  • 6% (1)
    I can hear the difference between D and E
  • 13% (2)
    I cannot hear the difference between D and E
  • 0% (0)
    I can hear the difference between D and F
  • 33% (5)
    I cannot hear the difference between D and F
15 Total Votes  
post #1 of 14
Thread Starter 

Here is a set of FLAC files for ABX testing, that compares the line output of onboard audio (Realtek ALC887) and Xonar D1 to a version of the original audio that has been upsampled with high quality software sample rate conversion. There are two tracks that can be used for testing, both were taken from previous blind test threads. The sound card outputs were recorded with the Xonar D1 at 96 kHz/24 bit (not an ideal solution, but should still show if there are any obvious differences), which has an input impedance of about 3800 Ohm. Onboard audio was tested at the maximum level, and the Xonar D1 at -2 dB to work around a limitation of the ADC. Some interference from the GPU was also created by moving a window on the screen during playback. The FLAC files are level matched, synchronized (not perfectly, but to less than one sample), and all are in 96 kHz/24 bit format.

 

A.flac - original (from here)

B.flac - onboard

C.flac - Xonar D1

D.flac - original (from here)

E.flac - onboard

F.flac - Xonar D1

 

post #2 of 14
Thread Starter 

Well, so far there is one vote for an audible difference (I wonder exactly what ?) with the onboard sound, and no difference with the Xonar card. Anyone else ?

 

post #3 of 14
How is this blind, when you've labeled the files?
post #4 of 14
Thread Starter 
Quote:
Originally Posted by skamp View Post

How is this blind, when you've labeled the files?


It becomes blind when using an ABX comparator normal_smile%20.gif I did not see much point trying to make it a secret which file is which, since in the previous blind test threads it was usually revealed quickly by others, and finding it out with an audio editor is also quite easy. However, in an ABX test, you can know exactly what A and B are, and it still shows whether you can hear the difference.

 


Edited by stv014 - 3/22/12 at 11:07am
post #5 of 14
Can't hear a damn thing. Looks like I didn't really need the Wolfson DAC of my FiiO E7, though I will continue to enjoy my placebo smily_headphones1.gif
post #6 of 14
The samples are unfortunate for a ABX test at under 80dB noise and distortion, to have any chance of hearing some difference you need human voice recorded well at different distances and small natural sounds in the background not synthesized instruments or compressed music with tens of effects on it smily_headphones1.gif You also need to listen longer and feel the difference over time, a 30 second clip will not help you much.
post #7 of 14
Thread Starter 
Quote:
Originally Posted by eugenius View Post

The samples are unfortunate for a ABX test at under 80dB noise and distortion, to have any chance of hearing some difference you need human voice recorded well at different distances and small natural sounds in the background not synthesized instruments or compressed music with tens of effects on it smily_headphones1.gif

 

Well, if anyone can suggest good test samples to use, I will create another set of files.

 

Quote:
Originally Posted by eugenius View Post

You also need to listen longer and feel the difference over time, a 30 second clip will not help you much.


This is a common argument, but ABX testing actually works better when comparing short segments of audio of at most a few seconds. The advantage of longer files is that there is more chance of finding a section where the difference is audible. But the total length of the samples is about 110 seconds, so it is not that short.

 

post #8 of 14
Not many takers, uh? Who would want to find out that they spent way too much money for nothing…
BTW, if anyone hears a difference, please post your ABX logs.
Edited by skamp - 3/24/12 at 10:20pm
post #9 of 14
Thread Starter 

Well, so far no one has been able to tell apart the Xonar D1 loopback (which also uses a minimum phase DAC filter as opposed to the linear phase one in the high quality "software DAC", for those who are interested). For the Realtek, the results are mixed. But the sample size is quite small, it seems this type of test is not very popular.

post #10 of 14

Only downloaded D and F, I can't hear any difference between the two.

 

Is Realtek onboard sound just really good when compared to other onboard sounds? Because all you ever hear is how "onboard audio sucks".

 

I have the ALC888 chipset, when I got my Fiio E10 I did a comparison between the two and while I could hear the difference it wasn't anything major like night and day. It was rather subtle difference of E10 being a tad warmer than the onboard sound. Though my experiment wasn't really that scientific and in the time that it took me to change the source I most likely forgot it all and my brain filled in the blanks (thanks to auditory memory lasting only a few seconds).

 

I'm yet to test this, but I think using just an amp and onboard sound should be pretty satisfying. Meaning that you could skip the separate DAC. I was thinking of getting something like the O2 in the future, which is really transparent to test this. Atm I'm little bit too low on cash to buy it just for testing this :p

 

I guess hearing is just so subjective and can be influenced too easily, to get any real conclusive results.


Edited by Papander - 4/26/12 at 8:34am
post #11 of 14
Thread Starter 
Quote:

Originally Posted by Papander View Post

 

Is Realtek onboard sound just really good when compared to other onboard sounds? Because all you ever hear is how "onboard audio sucks".

 

The newer onboard codec chips are not that bad actually, some (e.g. ALC889, ALC898, CS4207) could even be decent if implemented properly, although on most motherboards that is not the case. For example, with the ALC887 -> Xonar D1 loop I used in this test, the dynamic range is only ~89 dB (compared to 97 in the Realtek datasheet, or 110 dB with a loopback from the Xonar output), and it is even lower when there is system - especially GPU - activity. Somewhat oddly, high CPU usage can reduce then noise by 1-2 dB.

 

Edit: the noise problems noted above are at least partly the result of using a sound card for testing, which degrades the performance due to grounding issues. With the much noisier onboard ADC (90 dB SNR according to the specs), the dynamic range is still close to 89 dB, and is much less sensitive to system activity. Therefore, the noise is likely not as bad in reality as these tests - including the ABX files - show. The interference at about 2.9 kHz and its multiples on the graphs below is apparently also caused by ground loops, and disappears using the onboard ADC.

 

There are many reasons why onboard audio is generally thought to be poor:

- older generation chips were indeed bad, with low quality DAC filtering (rolled off and distorted treble), high distortion, and high noise

- even with decent chips, the actual implementation by the motherboard manufacturer is often less than perfect

- interference problems; this is variable, but can apparently be rather bad on some machines

- low quality headphone output (note that I tested line out) for a number of reasons: high output impedance (added by the motherboard manufacturer, for example, it is 77 Ohm on my machine, of which only 2 Ohm is in the ALC887), badly implemented front panel jacks with ground loops, and digital volume control, i.e. less than 100% volume makes the already mediocre dynamic range even worse

- various random issues related to buggy drivers and/or incorrect usage (driving headphones with a line output, bad mixer settings, etc.)

- subjective bias

 

Quote:

Originally Posted by Papander View Post

 

I'm yet to test this, but I think using just an amp and onboard sound should be pretty satisfying.

 

Adding a FiiO E11 or other decent but reasonably priced amplifier can be an improvement, indeed, as it makes the use of the digital volume control avoidable, and fixes any issues related to the quality of the headphone output.

 

I have created a few graphs of the ALC887 line output, all using 24 bit samples and maximum volume. Of course, with a sound card ADC, these are not 100% reliable, but may still be interesting to compare to some low end DACs like the FiiO D5. In general, the results are not bad, but the main problem is the high noise/limited dynamic range. Therefore, an onboard DAC is most usable with dynamically compressed music and/or loudspeakers.

 

Frequency response (44.1 kHz sample rate)    1 kHz sine wave, 0 dBFS, 44.1 kHz sample rate

 

20 Hz sine wave, 0 dBFS, 44.1 kHz sample rate    20 kHz sine wave, 0 dBFS, 44.1 kHz sample rate

 

20 kHz sine wave, 0 dBFS, 96 kHz sample rate    IMD test, 60 Hz 80%FS + 7 kHz 20%FS, 44.1 kHz sample rate

 

IMD test, 19 kHz 50%FS + 20 kHz 50%FS, 44.1 kHz sample rate    431 Hz -10 dBFS + 1201 Hz -90 dBFS, 44.1 kHz sample rate

 

Jitter test, 44.1 kHz sample rate


Edited by stv014 - 4/27/12 at 10:49am
post #12 of 14

Hmm I see I have voted earlier and couldn't hear the difference between A&B. Will test the rest of the files later.

 

I did some quick testing with my onboard audio ALC887. Actually I think it didn't sound that bad? But I still think I heard some kind of loss of fidelity.

 

Also the onboard audio for some reason was a lot quieter than the UCA-202. It was still loud enough to make my ears bleed though.

post #13 of 14
Thread Starter 
Quote:
Originally Posted by Headzone View Post
Also the onboard audio for some reason was a lot quieter than the UCA-202. It was still loud enough to make my ears bleed though.

 

Both have about 1.2 Vrms maximum output. But if you used them to drive low impedance headphones directly, then the actual maximum level can be lower due to high output impedance (47 Ohms for the UCA-202, and - at least with my ASUS motherboard - ~77, or more than 200 Ohms for the ALC887, depending on whether I use the headphone or line output).

post #14 of 14
Thread Starter 

Well, the files have been deleted by Mediafire. Maybe a new test could be created with different trakcs (suggestions ?) and better recording quality.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › BLIND TEST: onboard DAC