[voodoohao]

Hi guys, I've been reading up on DSD formatted audio files and there's both proponents and detractors of this technology (vs PCM - your normal 16/44.1 to 24/96 or 24/192). I have also heard that certain dacs (i.e Korg DSD Dac 100) are able to convert PCM files to DSD on the fly.

My questions would be:
1) If I use a good (relatively speaking) dac such as the Audiolab MDAC with 24/96 and above PCM files vs the Korg DSD Dac 100 upsampling function to convert those files to DSD, will the Audiolab sound better or vice versa?

2) if I listen to Youtube and Spotify predominantly instead of downloading music, which will be a better option? (Since the Audiolab's source will be mp3 quality songs while the Korg can still convert them upwards)

P.S the reasons why i picked these two dacs was because I can get them at used prices in my region.


Any thoughts and opinions are welcome!

 

[bigshot]

Upsampling won't make any difference at all.

 

[castleofargh]

Hi guys, I've been reading up on DSD formatted audio files and there's both proponents and detractors of this technology (vs PCM - your normal 16/44.1 to 24/96 or 24/192). I have also heard that certain dacs (i.e Korg DSD Dac 100) are able to convert PCM files to DSD on the fly.

My questions would be:
1) If I use a good (relatively speaking) dac such as the Audiolab MDAC with 24/96 and above PCM files vs the Korg DSD Dac 100 upsampling function to convert those files to DSD, will the Audiolab sound better or vice versa?

2) if I listen to Youtube and Spotify predominantly instead of downloading music, which will be a better option? (Since the Audiolab's source will be mp3 quality songs while the Korg can still convert them upwards)

P.S the reasons why i picked these two dacs was because I can get them at used prices in my region.


Any thoughts and opinions are welcome!

 
the short version is that DSD has problems, and PCM has problems, so once both have solved their own, they end up with about the same performances(hires PCM vs DSD). so close you kind of wonder what's the point of DSD in fact.
so whatever you can read, there is no benefit at all of having a master in DSD if you already have it in high resolution PCM. and that's not a subjective opinion on audibility, it's a technical fact (I have a great PDF somewhere explaining it all in great complicated details, I'll try to find it later).
 
now about the objective/audible perspective, it's been showed a few times now that people tend to fail at identifying the high res or the DSD from the 16/44 PCM in a blind test. that would make a few people like me say that high res is waste of space and money.
 
now specifically about your 2 questions:
1) you will gain nothing from a real time conversion from pcm to dsd or dsd to pcm.  that much is clear, you might actually end up with something of lesser quality depending on how it's done.
 
2) just like 1) you will gain nothing. mp3 becomes PCM for any DAC, but it doesn't magically turn into gold after being oversampled and converted to high resolution or to DSD. the data is still the one from the mp3. coca cola doesn't become wine because it's served in a wine glass. if you're gonna mostly use mp3/aac sources, don't bother with high res at all.
 
on the idea of upsampling/oversampling low resolution files, it isn't always advertised, but pretty much all modern DACs do it without telling you already. so really just buy something convenient that seems to go with your budget and don't think too much about it.

 

[limpidglitch]

 
on the idea of upsampling/oversampling low resolution files, it isn't always advertised, but pretty much all modern DACs do it without telling you already. so really just buy something convenient that seems to go with your budget and don't think too much about it.

 
Or alternatively, think about it a lot, and take a class in informatics/information theory.
It's a fascinating field, and applies to a lot more than just sound reproduction. I just wish I had the mathematical nous to understand it better.

 

[Strangelove424]

   
the short version is that DSD has problems, and PCM has problems, so once both have solved their own, they end up with about the same performances(hires PCM vs DSD). so close you kind of wonder what's the point of DSD in fact.
so whatever you can read, there is no benefit at all of having a master in DSD if you already have it in high resolution PCM. and that's not a subjective opinion on audibility, it's a technical fact (I have a great PDF somewhere explaining it all in great complicated details, I'll try to find it later).
 
now about the objective/audible perspective, it's been showed a few times now that people tend to fail at identifying the high res or the DSD from the 16/44 PCM in a blind test. that would make a few people like me say that high res is waste of space and money.
 
now specifically about your 2 questions:
1) you will gain nothing from a real time conversion from pcm to dsd or dsd to pcm.  that much is clear, you might actually end up with something of lesser quality depending on how it's done.
 
2) just like 1) you will gain nothing. mp3 becomes PCM for any DAC, but it doesn't magically turn into gold after being oversampled and converted to high resolution or to DSD. the data is still the one from the mp3. coca cola doesn't become wine because it's served in a wine glass. if you're gonna mostly use mp3/aac sources, don't bother with high res at all.
 
on the idea of upsampling/oversampling low resolution files, it isn't always advertised, but pretty much all modern DACs do it without telling you already. so really just buy something convenient that seems to go with your budget and don't think too much about it.

 
That's my impression too. I always thought PCM was just a raw bit stream that hadn't been packaged into a codec yet. I have no experience with DSD, but can't imagine how you could re-package PCM bit stream to improve the quality. It should be the same exact bit formation, any extra data is just filler, and any less would require a codec. I never understood why PCM was better than a packaged codec anyway. Sounds the same to me once the DAC gets a hold of either one. The only major benefit to PCM I found is universal compatibility. With one foot in PC systems and another foot in Mac systems, I take advantage of that quite a bit. 

 

[castleofargh]

 

 
on the idea of upsampling/oversampling low resolution files, it isn't always advertised, but pretty much all modern DACs do it without telling you already. so really just buy something convenient that seems to go with your budget and don't think too much about it.

 
Or alternatively, think about it a lot, and take a class in informatics/information theory.
It's a fascinating field, and applies to a lot more than just sound reproduction. I just wish I had the mathematical nous to understand it better.

hehe, sure I'm not saying it isn't interresting or vastly important as a process(that I also fail to fully understand). I'm more talking about how when sound can benefit from upsampling/oversampling it usually is already done anyway as NOS DACs are a thing of the past(or should be ^_^).

  That's my impression too. I always thought PCM was just a raw bit stream that hadn't been packaged into a codec yet. I have no experience with DSD, but can't imagine how you could re-package PCM bit stream to improve the quality. It should be the same exact bit formation, any extra data is just filler, and any less would require a codec. I never understood why PCM was better than a packaged codec anyway. Sounds the same to me once the DAC gets a hold of either one. The only major benefit to PCM I found is universal compatibility. With one foot in PC systems and another foot in Mac systems, I take advantage of that quite a bit. 

well from what I understand, some DACs chips(the real D to A part) speak PCM, some speak DSD. so for foreign languages like AAC MP3 FLAC... it's only a matter of who's doing the transaltion ^_^. up till now all of those codecs were built with PCM in mind so it's obvious that they're more comfy with it. but it wouldn't be that hard I guess to get a new codec that speaks DSD as second language. it's just not what we use now.
so at the moment if I use MP3, it will turn into PCM anyway and deciding to output it to DSD would force a double translation. even if it doesn't end up with loss of signal quality, it looks to me like a waste to do that for no concrete benefit.
 
/!\ warning I have no idea if any of the following is true, I didn't actually find or understood enough to make sure of it all, it's just me going wild. I'd be happy if someone could either confirm, or show me the light

.
but for anything working with pulse modulated signals I couldn't really understand how it really was different for PCM and DSD. the little I understood was that for PCM the DAC gets a sample with a value equivalent to a voltage value, and then does whatever it has to do to get to the right voltage using the appropriate voltage impulses. for DSD the signal already give the timings for the impulses, but the result sure look the same.
then I thought I had it when I realized that DSD DACs were one bit and PCM DACs where 24 or 32bit. but even that doesn't seem to stand as there are DSD using several bits(I'm guessing to reduce the noise from using a stupid 1bit signal). and there also seem to be some PCM dacs that actually use less bits as they don't really need a lot for pulse modulated signals.
so the more I look into it, the more it seems to me like DSD is just a rip off of delta sigma that pretends to be different by writting down time values instead of amplitude. but sound being made of sine waves, both are always linked directly and say exactly the same thing.
/!\ end of warning!

 

[blades]

   
Or alternatively, think about it a lot, and take a class in informatics/information theory.
It's a fascinating field, and applies to a lot more than just sound reproduction. I just wish I had the mathematical nous to understand it better.

 
You don't need math.  Common sense should get you there.  If you have a 16/44.1 red book music file and you want to  upsample it, where do you get the data to add to the data that is there?  If you create it out of thin air how does the algorithm decide where zeros should go and where ones should go?  That should be enough to get you to understand that upsampling has no audible purpose.  It might be useful for some sort of compatibility match but you certainly won't hear any difference between a red book and an upsampled red book file.  In short, you can put lipstick on a pig but it is till a pig.

 

[germanium]

Virtually all current ADC's convert analog to DSD then immediately to PCM. Conversion back to analog follows the same path in reverse ending with DSD as the final format before conversion to analog. Almost all DAC's do this now days so the question is really moot now, though the conversion to DSD is at a higher bit rate in such DAC's than the consumer release DSD found in SACD players.

 

[castleofargh]

Virtually all current ADC's convert analog to DSD then immediately to PCM. Conversion back to analog follows the same path in reverse ending with DSD as the final format before conversion to analog. Almost all DAC's do this now days so the question is really moot now, though the conversion to DSD is at a higher bit rate in such DAC's than the consumer release DSD found in SACD players.

in one of the last rocky mountain audio vids there is one that's pretty much a giant DSD provider's add. they mostly tell what they did great and why we should believe. still in the middle of this there are a few guys who can't help but be completely honest, and one explained that their was like 3solutions to record dsd and that the one that almost everybody used was as you said going to pcm for ease of use as dsd offered very limited mixing and editing possibilities.
 
there is also another vid I liked (I do enjoy those vids a lot overall) where the speaker pretty much says "you want DSD, I'll sell it to you, but it's a waste" ^_^. and he explains a few stuff like the fact that even sony gave up and admitted it wasn't a good commercial support.