Yes, I know but I thought you sold it and have now another NOS DAC...
Upsampling
- Thread starter Alex S
- Start date
nostromo
pfm Member
I had a quick play with upsampling on my system yesterday. I changed the Audrivana output from PCM to DSD and then in the upsampling section specified DSD64.
As you may have noticed I don't really know what I'm doing, there are far too many options and things to setup for my tiny brain to figure out what they do or not. Anyway fumbling away with these settings, I found the sound to be very 'rolled off' at the top end and it was an instant dislike from my ears. If this is what DSD64 is supposed to sound like then I don't like it one bit.
So I went back to what I had before, PCM with settings like his:
Instantantly my ears preffered this. The top end is back as it should be there is no roll off and the sound is more open and the treble is back.
This simple try out of DSD64 has left with a distict impression that I don't want to go anywhere near DSD again. But like I keep stressing all this is beyond me and I don't have a clue what these parameters do, I just want something sensible.
If anybody can suggest something that may be an improvement on this which I may like even more I'm prone to give it another try and perhaps have an even more euphonic setup than I already have.
tuga
Legal Alien
I don't use Audirvana but if you want to try DSD avoid DSD64, though I don't think that a higher rate DSD will make a difference regarding the 'perceived' top end roll-off.
The top end with PCM sounds 'crisper' and more 'obvious' than with DSD, which sounds 'smoother', although they'll both probably measure flat; choosing one or the other is a matter of preference, although the best measured performance (lower distortion) is achieve with the latter.
My suggestion is that you go on the Audirvana forum if there is one and ask for help there regarding filter set-up.
Jim Audiomisc
pfm Member
I'm now wondering about how easy it would be to write a prog that lets the user draw the required frequency-response and it generates the coefficients for sox to do the filtering...
tuga
Legal Alien
Well, it's Winter time and the nights are long.
Jim Audiomisc
pfm Member
tuga
Legal Alien
Didn't know that's what they're called but I do remember them well. You can still get them, my children, now teenagers, had one.
tuga
Legal Alien
Some comments from Grimm's engineers:
Since almost all DA converters run at a different internal sample rate and bit depth than the original audio format (such as 44.1/16 for CD), digital signals need to be converted from the original format to the ‘native’ format of the actual DA element. This conversion process is called oversampling.
It includes a phase linear filter that preserves the audio band, but removes its ultrasonic ‘mirrors’.
All converter chips contain such filters.
Our investigation has revealed that these filters pose a threat to sound quality when they have insufficient ‘processing power’ to calculate the filter and its requantization at the required precision.
For instance, the filtering is usually executed in several simplified, cascaded steps, which results in a loss of audio quality details.
https://www.grimmaudio.com/publications/the-pure-nyquist-filters-of-the-mu1/
Since almost all DA converters run at a different internal sample rate and bit depth than the original audio format (such as 44.1/16 for CD), digital signals need to be converted from the original format to the ‘native’ format of the actual DA element. This conversion process is called oversampling.
It includes a phase linear filter that preserves the audio band, but removes its ultrasonic ‘mirrors’.
All converter chips contain such filters.
Our investigation has revealed that these filters pose a threat to sound quality when they have insufficient ‘processing power’ to calculate the filter and its requantization at the required precision.
For instance, the filtering is usually executed in several simplified, cascaded steps, which results in a loss of audio quality details.
https://www.grimmaudio.com/publications/the-pure-nyquist-filters-of-the-mu1/
Jim Audiomisc
pfm Member
Curous that an 'investigation' was needed to discover what is techincally called "the bleed'n obvious" *defined* by the wording of the assertion.
The 'solution' is also obvious. The system just needs to provide sufficient "processing power". Since the input information rate is finite, so is the required "processing power".
John Phillips
pfm Member
That looks like a really bad argument from a technical PoV. Using "insufficient 'processing power'" for a desired outcome is bad engineering. Plain and simple. You fix bad engineering by doing the engineering properly. By curing the disease. Not by applying some "sticking plaster" to alleviate its symptoms.
There's a lot of processing power available today (for which I would have kiiled in the late 1980s to early 1990s when I was doing some work in the field). And the engineering knowledge of how to do things properly in digital audio has been widely distributed via the professional journals. It's not something that's a closely guarded secret known only to a few equipment makers and gurus ... except when it comes to markering.
tuga
Legal Alien
Nice marketing words.
tuga
Legal Alien
To be fair I think they mean insufficient 'processing power' for better than basic/crude processing (interpolation?).
Interestingly the MU1 is only able to upsample to 24/192, probably due to insufficient 'processing power'...
gez
pfm Member
Why does anything outside the audio range matter?
This is one thing that I've never quite understood when DACs are measured (or filters compared). Apart from the potential for high levels of ultrasonic energy to cause issues with a (badly designed) amplifier circuit, or blow tweeters, I don't see how it matters what happens outside the scope of what humans can hear.
gez
pfm Member
Being caused by what though? Because that doesn't happen if you feed a speaker with such a resonant peak in the ultrasonic range with a normal analogue signal, (as evidenced by the literally tens of such speakers lab measurements in countless reviews over the years). Something else must be happening. So are you saying a DACs output has caused this to happen? or that it's happening in the digital domain somehow, because there's simply no mechansim in physics that I can think of that would cause it to happen in the analogue domain.One example of a tweeter oil-can resonance modulating into the audio range:
gez
pfm Member
Well I must be deaf in that case (entirely possible), as I've never been able to hear aliasing. I've tried many times switching between CD/DAC filters that are supposed to have aliasing product and those that are supposed to remove, or significantly suppress them, and I hear no difference what so ever between them. At least not with real music.
Has anybody actually ever done blind ABX testing to prove aliasing is audible?
tuga
Legal Alien
See here for DACs aliasing into the audio band:
Aliasing, imaging and upsampling a real example
https://tinyurl.com/bde7stxm
Jim Audiomisc
pfm Member
One example of a tweeter oil-can resonance modulating into the audio range:
Interesting and curious. The result being distortion generated at fin/N? How? And I suspect this is level dependent. i.e. the distortion/signal ratio rises when the HF signal levle is lowered? Wondering what 'storage mechanism' causes ths type of resonance to generate subharmonics.
That said, yes, I do find it strange when reveiw measureents that show > 20kHz resonances in speakers then seem to ignore them totally.