Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Manfred]

DSD to PCM Conversion: The standard value used by MC is Safe (24kHz @48db/Ocatave) : Options->Audio>Advanced-> Configure input Plug-In->DSD conversion.

My Devialet has its own algorithm to convert DSD to PCM (taken from 8.1 Firmware Manual)

" MATŪ DSD Core technology converts the DSD format into Expert internal native PCM 40 bits / 384 kHz format thanks to an optimized algorithm, limiting the amount of computational operations on the audio stream. Using only 15 bit perfect additions (no multiplication, no storage), the MATŪ algorithm performs a 128 taps, linear phase FIR (Finite Impulse Response) digital filter ensuring the conversion of the DSD format into internal native PCM format with truly outstanding performances"

Is it now better to use the "Safe" Value or the  recommended value=off using external Low-pass filter?

[dtc]

If you send the DSD signal directly to the Devialet without PCM conversion, then the Devialet either plays it as DSD or converts to PCM.  The MC filter is irrelevant in this case, as it only applies when MC does the DSD to PCM conversions.

If MC does the DSD to PCM conversion, then the MAT DSD Core technology conversion is irrelevant, since the Devialet is not involved in the conversion.

The MC 24 KHz@48dB/octave filter is a very conservative fillter that was implemented as much for hardware compatibility as for sonic issues.  The Sony standard was 50 KHz@48dB/octave. Unfortunately,  MC does not implement that. The MC higher frequency filters unfortunately are not step enough to fully cut off all the noise.

If you are going to use MC for DSD to PCM conversion, you can  set the low pass filter to off and set your own low pass filter in DSP Studio - Parametric Equalizer.  I do that with the Sony recommended filter.

[Manfred]

Thank you dtc!

[dtc]

Glad to help.

[RD James]

The Sony standard was 50 KHz@48dB/octave. Unfortunately,  MC does not implement that. The MC higher frequency filters unfortunately are not step enough to fully cut off all the noise.

If you are going to use MC for DSD to PCM conversion, you can  set the low pass filter to off and set your own low pass filter in DSP Studio - Parametric Equalizer.  I do that with the Sony recommended filter.

Even at 48dB/octave, a 50kHz filter is letting a lot of ultrasonic noise through to your amplifier/speakers.
I do agree that all of the filters should be updated to use 48dB/octave rather than 24dB/octave, but I don't recommend anything higher than 30kHz.
Even then, many albums have spurious tones in the 24-30kHz range, and playing signals outside of the range of human hearing is likely to introduce distortion inside the audible range since you are making the tweeter do more work unnecessarily.

[kr4]

The MC 24 KHz@48dB/octave filter is a very conservative fillter that was implemented as much for hardware compatibility as for sonic issues.  The Sony standard was 50 KHz@48dB/octave. Unfortunately,  MC does not implement that. The MC higher frequency filters unfortunately are not step enough to fully cut off all the noise.

AFAIK, there are three options for this filter (in addition to bypass) and they can be selected under Tools > Options > Audio > Advanced > Configure Input plug-in > JRiver DSD plug-in. 

[dtc]

AFAIK, there are three options for this filter (in addition to bypass) and they can be selected under Tools > Options > Audio > Advanced > Configure Input plug-in > JRiver DSD plug-in.

Yes, one at 24 Khz/48dB per octave, one at 30 KHz/24dB per octave and one at 50 Khz/24 dB per octave. The first one is effective, but many people think too conservative at 24 KHz.  The cutoff was set because a well known DAC at the time did not handle higher frequencies well. The second two do not have  steep enough slopes to fully filter out high frequency noise.  If the second two had 48dB/octave slopes, then the 3 options would work well and present the user with appropriate choices and both RD and I could be happy.

[whoareyou]

What would be the advantage to setting filter to > 24khz? 

Isn't that above what we can hear?

[dtc]

That is a very long and involved discussion. Yes, you cannot hear a sustained tone at that range. But many people think there is an effect from transients above 20K.  In addition, the higher cutoff moves the ringing from the filter to higher frequencies, above the sustained hearing range, which can also effect the sound.  It depends on your equipment and your hearing and your experience.

As I said, it is a long, involved discussion with many people have very strong opinions. That discussion is best had on a forum devoted to audio. But choose your forum carefully, since different forums have different prevailing views.

You can also just try it and see if you hear any difference between the settings.

[whoareyou]

That's a discussion I'm not qualified to have at anything but a very generic level, which might be a good thing.

But I do like to tinker, and listen to the results.  Plus with Covid lockdown I'm looking for stuff to do.

So for example, I set my own filter for Sony recommended 50khz with 48db/octave.  Since I'm bit of an idiot with this stuff, what should the q value be set to? 

Thanks

[arrowc]

I recorded analogue from SACD, these files still have that very high frequency. I had to replace two tweeters, even though they were wav files. My settings are now 23   48db/octave in Parametric Equalizer.

[dtc]

How did you obtain the analog signal?  The analog signal from a SACD player should not blow out tweeters, otherwise this would have been a common problem when SACD's were popular and it was not.   Can you explain more about what you did?

[arrowc]

I agree it should not be there, but it is. I recorded from a sony sacd player that had 6 rca out jacks into my computer using rme and sound forge 192 24. I go into the analyzer adjust the upper end to say 90k and I can see the high frequency.

[dtc]

What's the typical level at 20,30,40 and 50 KHz? And at 90 KHz?

[arrowc]

I tried two files one 2ch one 5.1  the 2ch peaked at 65k with a span of about 3k at bottom, of course that changes with volume   5.1 peaked at 50k  These are wave files ?/   Striate dsf peaked at about 90k with a way sharper peak.

[arrowc]

the span is about 30k not 3

[dtc]

If the volume peaked at 65K, 50K and 95K , there is something very wrong with your files.  It is hard to believe that a properly functioning SACD output those signals.  By 50K, the signal should be low and decreasing.

Here is an example of the filters on a DSD converted to PCM file that I did a while ago.  You can see that the 24dB filter is not step enough.  But the levels in those graphs are more typically of what you should see in the higher frequency ranges.

https://yabb.jriver.com/interact/index.php/topic,97740.msg675689.html#msg675689

[arrowc]

very interesting. something did seem wrong to me, Thanks for your insight!

[dtc]

Not sure what happened, but it seems you have some unusual files. Not surprised they blew out your tweeters. Filters to the rescue.