[ITZBITZ]

I don't think I've seen this posted anywhere, but I compared a loopback test with the Emu 1212m using RMAA alone with a loopback test using Foobar2000 w/ASIO output and RMAA to record only. The results are staggering to say the least.

Foobar2000 Alone

Foobar2000 vs. RMAA

Either I screwed up or these results are just awesome, particularly the stereo crosstalk numbers.

 

[Glassman]

in fact the results look weird, especially the IMD graph.. would be interesting to play and record using ASIO from some pro app like Wavelab and then analyse using RMAA.. I guess there are some problems recording with WDM while playing back using ASIO..

I think the signals somehow mix in software, E-MU has a bit of rolloff at 20kHz which is caused by analog section of the card, the results obtained using foobar seem flat, that indicates something is wrong with the measurements..

 

[lan]

I also think the results look strange.

What you want to say/do is compare ASIO or kernel streaming (bypass kmixer) vs. directsound or waveout (through kmixer)?

You should try to use the same program foobar and switch between ASIO and directsound output methods and use RMAA in record mode in each case.

 

[Wodgy]

Thanks for trying this, but I agree that the results may not necessarily be correct. It's really hard to say. The data is definitely interesting though.

In particular, the different numbers for stereo crosstalk seem to indicate that there is something wrong with the experimental methodology. It is hard to imagine that kMixer does anything to the stereo separation. However, I suppose it is possible that kMixer introduces common distortion to both channels, beginning at about 1000 Hz, and this common distortion is being detected by RMAA as stereo crosstalk. If we accept this explanation, this would seem to indicate that kMixer only has about 84dB of correct dynamic range (reading off the stereo crosstalk graph). For processing 24-bit signals, that seems extraordinarily poor, even for something as broken as kMixer.

As Glassman pointed out, the distortion measurements are also hard to explain.