[stv014]

Here and here are some audio files recorded from the output of various low end (including some very old) PC sound cards, while driving 250 Ohm headphones near the maximum volume; in all cases, the actual voltage on the headphones would have been in the range 0.93 to 0.99 Vrms with a 1 kHz 0 dBFS sine wave. The recording and editing format was 96 kHz/24-bit, but this was converted to 48 kHz/16-bit at the end to reduce the file sizes. The files are level matched, channel imbalance is eliminated, too (although it was always well under 1 dB to begin with), and sample rate conversion was applied where necessary to correct minor differences in the actual sample rates of the cards tested. One sample was processed with a FIR filter to equalize the effects of a low damping factor on the frequency response. The original files (borrowed from these threads) are included as well, with the sample rate converted to 48 kHz for consistency. I do not tell what the files are yet to minimize bias, but a text file containing the solution has an sha1 sum of 9c614de5110f151995380d62626aa08db0f3c67e.
These are probably not difficult to tell apart, and some are even very easy. Post any interesting ABX results, or your opinion on what sounds best and/or worst.
 

 

[Swatcsi]

Thanks! I'll make sure to try this out!

 

[stv014]

By the way, I plan to create a similar test soon, that is simpler: it will only compare the recorded output of a newer "HD" onboard audio interface (Realtek ALC887) to the original. I wonder if anyone would fail to hear the difference

 

 

[stv014]

Well, there is not much interest, so here is what the files are:
 
A: old Sound Blaster Live Value card (emu10k1), hardware resampling to 48 kHz
B: original file resampled to 48 kHz
C: Asus Xonar D1
D: old Realtek onboard AC97 audio (ALC850), with software resampling to 48 kHz
E: Asus Xonar D1 with FIR equalization filter applied to the recorded audio
 
C (0.5 dB bass boost due to low damping factor) and D (audibly rolled off treble) vs. B is rather easy, and I heard the difference in all attempts. I would expect A and especially E to be somewhat more challenging, but I did not try testing these. Also, newer Realtek chips would not be so easy, this will be tested soon.
For completeness, here is a table of the maximum output voltage (without clipping) and the output impedance measured for each device, with the actual testing voltage used shown in parentheses (these are unloaded, so the real value is lower):
 
Realtek AC97:  0.987 Vrms, 13 Ohm
Creative SB Live Value:  1.401 (0.994) Vrms, 2 Ohm
ASUS Xonar D1:  1.935 (1.369) Vrms, 100.5 Ohm