The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Launch of Pono Studio Headphones
Old 5th May 2014
  #4711
Here for the gear
 
North Star's Avatar
 

Neil (Young); is he a member here? ...If not would you like me to invite him?
Old 5th May 2014
  #4712
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by Timesaver800W View Post
man have you ever heard any of these converters? it's like there's a 30 year delay on your postings.. bye-bye
Yikes. You might try listening to the man. He's not wrong. All DSD is, is insanely high sample rate PCM with next to no resolution. It weights error differently from standard PCM, in that all the garbage goes up to superhigh frequencies and accuracy increases with the lowness of frequency. I can do that with PCM dither and have, in a shipping product, so you needn't act like they are such different things (I even got a similar problem with too much high frequency noise, but by God could it resolve a low noisefloor at 20 hz! woot )

I understand it's hard to tell who's who in this enormous mega-conversation, but j_j's probably made the most sense of any of us. I could be persuaded that his estimate of 20/64K as an optimal delivery format is the best (no point, though, if we can jump to 24/96K on existing audio gear fairly easily). You might be surprised at the audio pedigrees of the many posters here who are seeing you tee off on JJ and sort of quietly shaking our heads in dismay. You're really hurting yourself more than anyone, by that attitude.
Old 5th May 2014
  #4713
Quote:
Originally Posted by North Star View Post
Neil (Young); is he a member here? ...If not would you like me to invite him?
I don't know and hell yes!
Old 6th May 2014
  #4714
Lives for gear
 

Quote:
Originally Posted by chrisj View Post
Yikes. You might try listening to the man. He's not wrong. All DSD is, is insanely high sample rate PCM with next to no resolution. It weights error differently from standard PCM, in that all the garbage goes up to superhigh frequencies and accuracy increases with the lowness of frequency. I can do that with PCM dither and have, in a shipping product, so you needn't act like they are such different things (I even got a similar problem with too much high frequency noise, but by God could it resolve a low noisefloor at 20 hz! woot )

I understand it's hard to tell who's who in this enormous mega-conversation, but j_j's probably made the most sense of any of us. I could be persuaded that his estimate of 20/64K as an optimal delivery format is the best (no point, though, if we can jump to 24/96K on existing audio gear fairly easily). You might be surprised at the audio pedigrees of the many posters here who are seeing you tee off on JJ and sort of quietly shaking our heads in dismay. You're really hurting yourself more than anyone, by that attitude.
there's a reason people use different words for different things, i could just call it green what other people call blue and yellow, doesn't make things very specific in my opinion.

you could just as well call pcm dsd cause of the delta sigma in many converters. shake your head in dismay as much as you like. i have zero respect for most of the participants here anyway, nil.
Old 6th May 2014
  #4715
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by Timesaver800W View Post
you could just as well call pcm dsd cause of the delta sigma in many converters. shake your head in dismay as much as you like. i have zero respect for most of the participants here anyway, nil.
Then we disagree, because I have plenty for many posters on all sides of the argument, some of whom I've seen literally shake off these types of hostilities and communicate as friends. If Neil Young DOES turn up here, I hope to hell we still have folks who remember how to do that. Otherwise, I feel really sorry for him because as much as I respect his ear and his agenda in this, I don't think he can keep up technically with this thread.

Experimenting with the Average plugin I made for the thread, it occurs to me that the sound it makes from using a lot of average taps is very much like backing away from the sound a long distance in open air. Just a thought on possible reasons why they're picking certain techniques for their converter…
Old 6th May 2014
  #4716
Lives for gear
 
nuthinupmysleeve's Avatar
 

Quote:
Originally Posted by Timesaver800W View Post
i have zero respect for most of the participants here anyway, nil.
...and why is that?
Old 6th May 2014
  #4717
j_j
Lives for gear
Quote:
Originally Posted by Sounds Great View Post
Ok, so how does that one bit PCM sound?
Just like SACD because that IS 1-bit PCM.

Next, please?
Old 6th May 2014
  #4718
j_j
Lives for gear
Quote:
Originally Posted by Sounds Great View Post
Well you haven't really done that, except to state that it is so.
I've pointed you to a deck that explains the technology. I am not responsible for your unwillingness to do basic research into the most simple, trivial fundamentals of the technology you appear to use.
Old 6th May 2014
  #4719
j_j
Lives for gear
Quote:
Originally Posted by Timesaver800W View Post
you could just as well call pcm dsd cause of the delta sigma in many converters. shake your head in dismay as much as you like. i have zero respect for most of the participants here anyway, nil.
Your reasoning in the first half of this failure to engage is a straw man. That is not why DSD and SACD are forms of PCM. They are forms of PCM because they encode things in a PULSE COD MODULATION, which is what PCM stands for. There is no decoder, so it's not DPCM. There is no adaptive step size, so it is not APCM. Obviously, it's not ADPCM because of the lack of both adaptivity in the step size and lack of predictor.

It's pcm. Heavily noise-shaped, highly oversampled PCM, indeed, but one can do both noise shaping and oversampling of many different kinds at any level of oversampling or noise shaping within a very broad envelope, starting with baseband PCM and noise shaping that is similar to absolute thresholds of hearing, and moving on to high oversampling rates and high order noise shaping, indeed as SACD and DSD make clear by themselves.

You may have no respect for those here who point out that you are being willfully disputatious, and have offered neither reasoning nor a testable premise for your claims, accusations, and insinuations. As such, this makes you, in my opinion, a disruptor whose effect is to muddy the waters and obscure the actual science and mathematics involved in audio.

You have made many untestable claims. How about you show some actual testable, verifiable evidence for your claims. Put your evidence on the line, and we'll see how it stacks up.

Your unwillingness to actually engage speaks to something, but not something positive, about your position.
Old 6th May 2014
  #4720
j_j
Lives for gear
Quote:
Originally Posted by chrisj View Post
Experimenting with the Average plugin I made for the thread, it occurs to me that the sound it makes from using a lot of average taps is very much like backing away from the sound a long distance in open air. Just a thought on possible reasons why they're picking certain techniques for their converter…
Why don't you make a plug-in with a better filter? It would require weighting before you take your "moving average", nothing more.

The reason that using the moving average sounds like "backing away" is that you're cutting high frequencies, which do not transmit as well through humid, room temperature air as low frequencies. If you want a set of filter taps, tell me the sampling rate and length, and I'll give you some filters to try out.

Such effects may be euphonic, but they are not accurate.

I think that euphonic distortions (guitar amplifiers are a classic example of that) may be the cause of much of this dispute. Tape scrape-flutter, the various interchannel IM distortions of LP's, the loudness growth due to distortion on both tape and LP's, etc, may partially offset the huge loss of information that happens when we reduce a soundfield to 2 points of reproduction, at least as far as the ear is concerned. Certainly some such effects are easily documented.
Old 6th May 2014
  #4721
Here for the gear
 

hello j_j
nice to meet you (virtually)
but you can play with the words indefinitely: the 2 principles are different...
I'm not saying that DSD is better by nature (probably even DxD can be greater?) and maybe they are complementary; but there's a so enormous complexity/energy for equivalent result (1GB for 10mn stereo DxD: imagine with 48 tracks!)... I repet but why don't let a chance to DSD?
In Darwin's theorem (dearly at scientists) when 2 processus are in conflict, the more efficient will win at long term; what's happen if there's no choice?

I am not an AES expert and probably never will be, but scientists don't have 'absolute knowledge'; before Copernic, the earth was motionless for everybody even for other scientists!

But I am a pragmatic listener, if there's a chance/possibility of true improvement, why don't let any chance...?

Musically
Old 6th May 2014
  #4722
Lives for gear
 

Does a PCM converter (16-24/44.1-192) utilize a decimation filter for AD, an interpolation filter for DA? Yes or no?

Does a DSD converter utilize a decimation filter for AD, an interpolation filter for DA? Yes or no?

Are there a bunch of guys who prefer their Korg DSDs to their Lavry Golds over in the high end section of this forum, as we speak? Yes or no?

if one must call DSD something different than DSD, call it PDM.
Old 6th May 2014
  #4723
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by j_j View Post
I think that euphonic distortions (guitar amplifiers are a classic example of that) may be the cause of much of this dispute. Tape scrape-flutter, the various interchannel IM distortions of LP's, the loudness growth due to distortion on both tape and LP's, etc, may partially offset the huge loss of information that happens when we reduce a soundfield to 2 points of reproduction, at least as far as the ear is concerned. Certainly some such effects are easily documented.
Absolutely. I was talked into including flutter on 'ToTape' only to discover it brought a particular kind of spatiality associated with tape recordings, and loudness boosts through continuously increasing distortion is a given these days. I go back and forth between using my fake-tape and not using it, knowing that it brings me something but also having a weakness for the sheer accuracy of a minimal mix buss.

I do have a plugin, a freebie no less, which can do your 'better filter' at least to an extent. It's Toybox, which is just a set of sliders corresponding to FIR taps (the window's probably too small for what you're interested in). The point I'm making (or at least considering) is, if this moving average corresponds so closely to the fall-off of treble through air, surely we are going to hear it as more natural than a treble filter that doesn't correspond to the sound of physical distance through air?
Old 6th May 2014
  #4724
Lives for gear
 

Quote:
Originally Posted by chrisj View Post
... The point I'm making (or at least considering) is, if this moving average corresponds so closely to the fall-off of treble through air, surely we are going to hear it as more natural than a treble filter that doesn't correspond to the sound of physical distance through air?
Surely we are not.

The distance from the source to the microphone determines the initial rolloff. This is set by the recording engineer to produce the "right" sound, usually by listening to the direct output of the desk before recording / digitisation. On final reproduction, the challenge is to bring this same sound to the listener's ears without any further modification. Adding artificial distance cues may produce a pleasing sound, but it isn't quite what the artist / engineer / producer intended you to hear.
Old 6th May 2014
  #4725
j_j
Lives for gear
Quote:
Originally Posted by AudioTouch View Post
But I am a pragmatic listener, if there's a chance of true improvement, why don't let any chance...?
Is there some reason you are suggesting I am not "let any chance"? I'm just trying to get people to understand what the technologies they are arguing about actually are.
Old 6th May 2014
  #4726
j_j
Lives for gear
Quote:
Originally Posted by Timesaver800W View Post
Does a PCM converter (16-24/44.1-192) utilize a decimation filter for AD, an interpolation filter for DA?
Who's talking about a convertor? Baseband PCM is baseband PCM.

There are antialiasing and antiimaging filters in all PCM systems, including those in SACD and DSD. Certainly those in highly oversampled systems should not impact the actual 20-20k or thereabouts range, but don't forget the 50kHz (if I recall correctly) 5th order chebychev? spec'ed in SACD, for instance.

I'm not quite sure why you are changing the subject now, so perhaps you could explain why you think a rate conversion might be necessary in baseband PCM, which is the original context of the "what filter" question I asked you. If you're doing rate conversion, it's not baseband by definition.
Old 6th May 2014
  #4727
Here for the gear
 

Quote:
Originally Posted by j_j View Post
Is there some reason you are suggesting I am not "let any chance"? I'm just trying to get people to understand what the technologies they are arguing about actually are.
the last question is not personnal; it's just a reflection about the 'why and how' the DSD format isn't more developped?!?

But part of the answer is given by the team PonoMusic when asked about the DSD : "it simply Does not Have broad enough acceptance by Consumers, studios or labels." Basically, it is not asked we do not offer it...

OK, so it is proposed to increase the resolution and the frequency of a PCM format keeping all its faults because it does not bother too!

But we do not see all benefits to be gained with the arrival of this format (DSD) which is much closer analog...

Musically
Old 6th May 2014
  #4728
Gear Guru
 
Sounds Great's Avatar
 

Quote:
Originally Posted by j_j View Post

There are antialiasing and antiimaging filters in all PCM systems, including those in SACD and DSD.
Why would you say SACD AND DSD?
Old 6th May 2014
  #4729
j_j
Lives for gear
Quote:
Originally Posted by Sounds Great View Post
Why would you say SACD AND DSD?
Why do you think, now? One is multibit ovesampled PCM, one is single bit oversampled PCM. Two different examples.
Old 7th May 2014
  #4730
j_j
Lives for gear
Quote:
Originally Posted by AudioTouch View Post
the last question is not personnal; it's just a reflection about the 'why and how' the DSD format isn't more developped?!?
One simple reason. DSD is nearly impossible to process, as it requires much more memory, many more MFLOPS, *and* much deeper bit depth in the processing in order to be able to do anything useful.

In fact, most processing of DSD is done on DSD reduced to 384kHz high-bit-depth PCM, and then reprocessed back up, and even 384kHz causes problems with EQ and filter coefficient mantissa length (or bit depth for fixed point system).

It was, some people concluded, an attempt at 'security by obscurity', in that it requires a much higher storage and streaming rate and it's agonizing to process. I can't speak to intent, but the effect on processing is certainly a reasonable statement.
Old 7th May 2014
  #4732
Gear Guru
 
Sounds Great's Avatar
 

These are nothing more than a continuation of the same discussions we are having here, with the same bias.

--


Quote:
The differences between the two processes are relatively minimal and somewhat akin to comparing 10 dimes with a 1-dollar bill and trying to determine which is better. In one case where you have a coin purse, 10 dimes is better. In the case where you have a wallet though, the dollar bill is preferable. In either case, you still have the exact same amount of currency. Likewise in the case of PCM vs. DSD, you still have the same amount of data being processed. It's simply a matter of how the decoding hardware (DAC) was constructed as to which format is preferable.



Quote:
In fact, during the recording process virtually all DSD recording are first converted into PCM for mixing, equalization and other "artistic" processing. Upon completion, the data is then converted back to DSD format and manufactured into discs of "so called" DSD recordings (SACD) for distribution to consumers.
Unless they're not.

Quote:
In addition, the majority of DSD recordings are actually converted from original PCM masters.
Unless they're not.

Quote:
2) Since DSD is a one-bit format it is literally impossible to perform any signal manipulation at all — even a fade-out. So to perform recording in the modern methods where signals are mixed, EQ’d faded, reverb added, et cetera, all of the DSD signals must first be transcoded into PCM (or analog) signals, then the signal processing applied, and finally re-modulating the signal back into DSD, adding another layer of high-frequency noise.
Uh, ok. Kind of like records, or tape? So this has what to do with the price of tea in China?

Quote:
3) Except for the microphones, amplifiers, and loudspeakers, all of the equipment in both the recording and playback chain must be replaced with new hardware that can accommodate this new form of modulation.
Ya think?

Quote:
We can see that DSD has quite a few strikes against from the very start. So the only justification for it whatsoever would be for the hard-core audiophile, and this would only be in the case if DSD could provide audibly superior sound quality over PCM.
Bingo. So, the big question is...
Old 7th May 2014
  #4733
Lives for gear
Quote:
Originally Posted by j_j View Post
Why do you think, now? One is multibit ovesampled PCM, one is single bit oversampled PCM. Two different examples.
This is not true.
Old 7th May 2014
  #4734
Here for the gear
 

Quote:
Originally Posted by j_j View Post
One simple reason. DSD is nearly impossible to process
that's true with actual knowledge; if there's no research on this point it'll be always the case...

Quote:
Originally Posted by j_j View Post
as it requires much more memory, many more MFLOPS, *and* much deeper bit depth in the processing in order to be able to do anything useful.
I'm not agree, if you compare what is comparable (DXD), the DSD format is 4x lighter and lossless compression DST MPEG4 is between 8 and 12 time lighter (maybe improvable if there's reseach)...
We can debate about it but for me the PDM (DSD&DST) is more efficient that equivalent discretisation 'PCM' - I speak only for Audio domain

Quote:
Originally Posted by j_j View Post
In fact, most processing of DSD is done on DSD reduced to 384kHz high-bit-depth PCM, and then reprocessed back up, and even 384kHz causes problems with EQ and filter coefficient mantissa length (or bit depth for fixed point system).
this is the method used by Merging/Pyramix with DXD (352.8 kHz/24bit)
That's why I said the both formats could be complementary;

If you want real time treatment, you can always use analog process...

Musically
Old 7th May 2014
  #4735
j_j
Lives for gear
Quote:
Originally Posted by Sounds Great View Post
And an SACD is merely a disk (standard) that is storing 1 bit DSD sound, so please explain to me once again how your are comparing DSD to SACD.
Actually, there are two formats, the pro (multibit) format and the 1 bit format.

And the 1-bit format is what you get on an SACD.

Many people refer to the pro format as DSD. Maybe they are wrong. I don't know, frankly. The point holds that there are both single bit and multibit systems running with the same order of noise shaping filter at the same sampling (but not bit) rate.
Old 7th May 2014
  #4736
j_j
Lives for gear
Quote:
Originally Posted by MarsBot View Post
This is not true.
As is common in these discussions, you controvert without saying specifically what you are controverting.

Are you disagreeing that the format on the physical SACD is 1-bit oversampled PCM?

Are you disagreeing that the pro format is multibit oversampled PCM?

What are you disagreeing with?

If you would like to call the multibit system something else, fine. I've seen usage all over the map.
Old 7th May 2014
  #4737
j_j
Lives for gear
Quote:
Originally Posted by AudioTouch View Post
We can debate about it but for me the PDM (DSD&DST) is more efficient that equivalent discretisation 'PCM' - I speak only for Audio domain
Perhaps it's a language difficulty, but if you're using the "PDM" title for "pulse density modulation" like some folks have tried to make DSD/SACD out to be, well, that's just a marketing hedge. Yes, I know, lots of people have said the same thing. No matter how you cut it, DSD is just oversampled, noise-shaped PCM.

Consider: PCM means Pulse Coded Modulation.

DSD is coding pulses. It is PCM. It requires no further modification, because like any other form of PCM, you can put it directly into a reconstruction fillter (which I cheerfully agree is very simple for DSD) without any need for any kind of decoder. People can make up all the names they want for it, it's still oversampled, noise-shaped PCM.

Note, it is trivial to come up with forms of "PDM" that do not have the same results as DSD, and do not act like a linear system as DSD does, within its intended bandwidth. This should be a further hint that PDM is simply not an appropriate name for it, really.

As to the rest of the sentence, I can't figure out what you're saying. DSD discretizes audio like PCM because it is PCM. It uses a different exact form of discretization because it runs at a different rate with different bit depth and different noise-shaping and dithering (well, lack of dithering in 1 bit systems, which is not a biggie in this particular case) systems.

PCM is PCM, oversampled, noise shaped, or not. PCM is a broad class of systems that are used in many different ways in different applications. There are many forms inside of modern PCM convertors (yes, some of which are a tad lame, yes, we've all noticed, thank you).

Oh, and, processing DSD of any form is much more complex than processing standard high-bit-depth PCM at lower sampling rates. The issue is the sampling rate, if you want to do any kind of EQ or filtering, while the bit depth does matter, the actual coefficient length is controlled primarily by the sampling rate when the frequency of interest is very, very much smaller than the sampling rate. Rabiner and Gold has a decent discussion of this, even if it is very early DSP book. This is not a case of "further research", it's fundamental mathematics.

That's for IIR filters. For FIR filters, the length requires enormous calculation depth (as well as MFLOPS) to avoid noise buildup. In any case, DSD is very hard to process compared to PCM, which is a completely germane comparison. Even 192kHz causes very much less problems than DSD, which is probably obvious from the ratio of the two sampling rates.

Oh, and "efficiency" is a mathematical thing in that baseband PCM is more efficient in information-theoretic terms, as well, this of course when presuming one is encoding a specific baseband via whatever method. That comes down to subtractive vs. multiplicative reduction in error per bit. But now we are getting WAY esoteric in audio terms, I think.

As to Pono, frankly, I have no idea how it will work out. Maybe it will be an excuse to stop using a compressor set to infinity on everything. Please? Pretty Please?
Old 7th May 2014
  #4738
Gear Addict
 

Quote:
Originally Posted by Sounds Great View Post
These are nothing more than a continuation of the same discussions we are having here, with the same bias.
So, the big question is...
"Why doesn't Pono get behind DSD?" Am I close?
Old 7th May 2014
  #4739
Here for the gear
 

Quote:
Originally Posted by j_j View Post
Perhaps it's a language difficulty,
yes it's a language difficulty, abuse even:

wiki source:"The word pulse in the term Pulse-Code Modulation refers to the "pulses" to be found in the transmission line. This perhaps is a natural consequence of this technique having evolved alongside two analog methods, pulse width modulation and pulse position modulation, in which the information to be encoded is in fact represented by discrete signal pulses of varying width or position, respectively. In this respect, PCM bears little resemblance to these other forms of signal encoding, except that all can be used in time division multiplexing, and the numbers of the PCM codes are represented as electrical pulses."

for me, the words have a limited interest; tell it like you want...

but physically it's not hard to understand that the principles are quite different:
one is representation of amplitude at a given time (discretisation)
the other just indicate if it's more or less (1 or 0) : whence 1-bit coding (DSD/DST/PDM...?)

Musically
Old 7th May 2014
  #4740
Gear Head
 
bandpass's Avatar
 

Quote:
Originally Posted by AudioTouch View Post
but physically it's not hard to understand that the principles are quite different:
one is representation of amplitude at a given time (discretisation)
the other just indicate if it's more or less (1 or 0) : whence 1-bit coding (DSD/DST/PDM...?)

Musically
No, as JJ said, think about how they're reconstructed: in all cases, a quantised voltage (be it to 2, 5, 65536, or any number of levels) is LPF'd to give the analog output. The principles are exactly the same.
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump