INTERACT FORUM
Devices => Sound Cards, DAC's, Receivers, Speakers, and Headphones => Topic started by: InflatableMouse on August 26, 2013, 04:06:30 pm
-
With DSD128 or 2xDSD, there should be no noise below 40 kHz, so a higher low pass filter is desirable than with DSD64. If 30 kHz was too high for 1xDSD, isn't 50 kHz too high for 2xDSD (I think 30 kHz would be too low for 2xDSD)?.
I also think I would prefer this process to be automatic. If 24 kHz is preferable for 1xDSD, would 2xDSD not be better served with a 44 kHz filter? Couldn't MC not determine the best filter based on 1x or 2xDSD? Next to that I would also like to disable the low pass filter entirely so I can add my own preference in PEQ.
Another Q related to this. My Teac UD-501 has 4 selectable FIR filter settings:
FIR1: fc = 185kHz, gain = -6.6dB
FIR2: fc = 90kHz, gain = +0.3dB (default)
FIR3: fc = 85kHz, gain = -1.5dB
FIR4: fc = 94kHz, gain = -3.3dB
When I bitstream DSD, I would bypass the low pass filter in MC, but these FIR filters are much higher. Wouldn't that potentially allow noise to come through and isn't that undesirable? When you look at this image (http://4.bp.blogspot.com/-B_XzJLefhYM/UY9N31sQmVI/AAAAAAAAA0g/4OBNd5ykp_E/s1600/Noise.png), it looks like a considerable amount of noise is allowed through. What am I to make of this?
When I'm not bitstreaming, apart from the gain difference, these FIR settings aren't doing anything due to the filter in MC, is that correct?
Thanks!
-
The low-pass only applies when converting to PCM.
If you're outputting DSD, the DAC must do the low-pass (or you're going to fry your tweeters).
-
For those of us who are converting to PCM, it does seem like 2xDSD (or higher) would require a higher low-pass filter to avoid filtering out any potential high frequency content.
I also wonder if the current 24kHz@48dB/octave filter is optimal for converting to PCM. In my limited testing using Media Center's conversion options, it seemed like you could use a higher frequency low-pass and still filter out most of the ultrasonic noise, providing greater frequency extension. (though the other two input filter options left a ton of noise)
I understand that you might not want to focus much effort on that right now, when MC19 is still in the early stages of development, and it could be argued that any changes would be inaudible anyway.
-
I believe InflatableMouse contacted Archimago about this. He asked me about it offline. I'm going to copy my reply here because it explains how to configure the lowpass and why we're using the filter we are:
I updated the wiki here with a little more detail about where to configure the low pass:
http://wiki.jriver.com/index.php/DSD_Format#Playback
Currently we don't allow picking different low pass settings for different samplerates of DSD. It's probably a good idea.
I need to try playing some 2xDSD samples and look in 'Analyzer' to see how the energy profile looks. I've found a lot of variability in 1xDSD (which is why we use a conservative lowpass now).
I linked to this in the wiki, but I'll do it again here as it explains a little about why we picked our default:
http://www.computeraudiophile.com/f11-software/software-conversion-dsd-and-2x-dsd-24-192-a-16871/#post242281
-
The fact that you believe its probably a good idea probably means you will think about it some time. That's good enough for me for now.
-
I ask politely: Any news? Is it possible with J.River22 to use different filters vor DSD, 2xDSD, ...