On Thu, 27 Jul 2017 07:41:19 -0500, Puffer Belly wrote:
> Good information to know. I enjoy DSD and hi-res PCM. There are a lot
> of DSD bashers and it's hard to tell what's fact vs. opinion.
>
> I do check hi-res audio with Audacity and have noticed many high
> sampling rate music rolls off sharply at 44.1 kHz (early digital
> recording or CD-master conversion) or it rolls off gently at 50 to 60
> kHz (tape). Does anyone know anything about Paul Simon's hi-res albums
> from the 70's and 80's? All roll off at 44.1 kHz even though they were
> recorded on tape and all are being sold as 96 kHz files.
>
>
>
> On Wed, 26 Jul 2017 18:50:42 -0500, SuperAudioLabs <Super@udio.labs>
> wrote:
>
>>On Mon, 24 Jul 2017 07:39:39 -0500, Puffer Belly wrote:
>>
>>> You didn't understand hardly anything I wrote. When DSD is converted
>>> to PCM, which is how people without DSD DACs listen to DSD, the
>>> equivalent sampling rate (FOR PCM NOT DSD) is a multiple of 44.1 kHz.
>>> As I stated, DSD64 (SACDs) are converted to 24 bits sampled at 88.2
>>> kHz, DSD128 176.4 kHz. etc. When people convert DSD to PCM using
>>> sampling rates other than what I just listed, such as a multiple of 48
>>> kHz (96, 192 kHz, etc.), that causes more distortion than using a
>>> multiple of 44.1 kHz. Converting DSD to PCM or vise versa is not
>>> distortion free. If the math confuses you, then learn something about
>>> converting DSD to PCM.
>>>
>>>
>>> On Sun, 23 Jul 2017 17:23:18 -0500, SuperAudioLabs <Super@udio.labs>
>>> wrote:
>>>
>>>>On Sun, 23 Jul 2017 11:31:11 -0500, Puffer Belly wrote:
>>>>
>>>>> I'd just like to point out that DSD sample rates are based on
>>>>> multiples of 44.1 kHz, so DSD64 is equivalent to PCM sampled at 88.2
>>>>> kHz, DSD128 176.4 kHz, etc.
>>>>>
>>>>>
>>>>That is actually entirely incorrect. That is just some modern bullshit
>>>>some anti-DSD guys came up with to bash the format.
>>>>
>>>>DSD is not the equivalent to anything PCM....the two are basically
>>>>like gasoline and diesel...similar products with an entirely different
>>>>level of operation.
>>>>
>>>>Equivalence in PCM is made up...much in the same way we make up a
>>>>horizontal resolution for analog video back in the day when there
>>>>really was no such idea.
>>>>
>>>>IF what you said had any kind of truth to it; than the CD players of
>>>>yester-year which worked with 2x, 4x, or 16x oversampling would not
>>>>have been able to produce the sound quality they did....as you
>>>>yourself said DSD64 is the equivalence of 44.1khz.
>>>>
>>>>BTW, any time you see something like "oversampling", or "MASH", of
>>>>"Pulse Edge Modulation" on a CD player...it means it's using a
>>>>delta-sigma 1-bit modulator internally. It's taking the 16-bit data
>>>>from CD and modulating it at 1-bit....at a much slower rate than DSD.
>>>>
>>>>DACs up till about 10 years ago were using 1-bit DS modulation; these
>>>>days the higher end models are using multi-bit delta-sigma.
>>>>
>>>>No one has made a DAC that can handle more than 16-bit samples that's
>>>>not an oversampling design.
>>>>
>>>>DSD64 just means it's sample rate is 64 times the normal CD rate. Why
>>>>was the CD sample rate held to a standard? Because it was the first
>>>>time they actually standardized a sample rate.
>>>>
>>>>Actual performance of any DSD rate depends on what created the DSD
>>>>stream and what is playing back the DSD stream.
>>
>>I took what you wrote literally. I was never talking about conversion,
>>you never mentioned conversion.
>>
>>I'll also add that the conversion technology has matured; 20 years ago
>>resampling from say 48khz to 44.1khz was pretty bad. These days, the
>>process is not a problem any any distortion you get is usually
>>inaudible.
>>
>>DSD conversion itself has matured a lot. Originally the only sample rate
>>ever used for DSD -> PCM was 352.8khz; which yes...does fold evenly in
>>to 44.1. But that was also close to 20 years ago (13 - 17). There has
>>been development in the DSD world; used to be you couldn't really do
>>*any* processing in DSD...now there is software out there that will let
>>you do basic equalization and level adjustments while in 1-bit.
>>
>>I'm pretty sure the software we're using here actually converts the DSD
>>to the standard "DXD" PCM rate...then further resamples using FFT based
>>SSRC.
>>
>>In 2002/3 I would have been right there with you; but the technology and
>>code has matured quite a bit.
>>
>>Plus...have you ever looked at some of the 192khz stuff from HDTracks?
>>They claim they give us what they get "from the record label"...in which
>>case, the record label is outputting stuff from DSD in the highest
>>consumer rate they can.
>>
>>Even worse...you ever seen a SACD that actually states it was sourced
>>from PCM? I seem to recall some of the SHM-SACD Queen releases even
>>stated they were made from 96/24 PCM.
>>
>>It's easy to tell a DSD->PCM conversion, it's next to impossible to tell
>>PCM->DSD. There's a SACD ISO of "Will The Circle Be Unbroken" floating
>>around that we sourced from the 192/24 HDTracks.
>>
>>Then there's my colleague next door from AHUSL who has been playing
>>around with feeding high-end ADCs with a direct feed from high-end DACs.
>>But I think he's just looking for excuses to play with his new reflow
>>station.
There are a lot of people who make claims about DSD...then try to use
scientific fact to back it up. I then ask them if they do a spectral
analysis to determine if a wine (or bourbon...vodka if you're dirty like
theAudioHut) is better. Usually you don't...you effing drink it and make
that determination.
Audio is the same way; the science can say everything it wants...in fact
it says a lot. Early science said digital audio wasn't possible, then it
was. Early science said "you'll never store bits on a disc and read them
with a laser"...then it was. If you'd told a couple guys back in the early
40s that a piece of germanium sliced and processed properly would replace
vacuum tubes....they'd have laughed at you and probably thrown 6L6 tubes
at you. (well, maybe 2A3s since those were "obsolete" then and 6L6 was
still in large use...which lead to the 2A3 shortage of the 90s and early
2000s. It's ok now...some tube factory is making them again).
Anyway, I think the problem is that SACD came out at the "wrong" time to
be taken for what it was. The idea wasn't necessarily to create some super
high resolution format with enough dynamic resolution to measure particle
movement; if that's what Sony's goal was they failed at it pretty badly.
Instead...one needs to look at when SACD came out, 1999, and realize it
was probably in development for a few years. It's goal was to solve a
major limitation that'd existed in DACs...and that is delta-sigma
modulation. A R2R-ladder DAC was very expensive and tedious to build, you
needed to have all of your resistors match if you wanted to have any kind
of standardized quality...and precision matched resistors are expensive.
So no one was really making R2R-ladder DACs, which work on pure PCM
samples, outside of very specialized niche uses. Pretty much everyone who
was making a CD player was using the "oversampling" method...which
converts the PCM in to a delta-sigma modulated bitstream; this 1-bit
stream is what is then actually made analog. I do believe a few players
used pure PCM DACs...and they were regarded as garbage. The 1-bit/
oversampling technique was "far superior"...and that's part of the reason
they started slapping things like "1-bit dual DA converters" and "whatever-
X oversampling" on players.
It was this conversion that was the major weak link. Part of the reason
the quality among players varied was due to how they implemented DS
modulation and all the internal filtering. DSD/1-bit requires noise
shaping...and a LOT of it. The early CD players were using 2 or 4x
oversampling, 8 and 16 came later. Much of the research that occurred was
in the noise-shaping and conversion process. Everyone had a trademark name
for what they'd come up with too; M.A.S.H. from (I think) Technics stands
out as they prominently stuck it on every CD player. MASH wasn't just that
stupid game girls played on the playground...or that show that had
Klinger...it was a special noise-shaping technique Technics (if I'm right)
came up with for the delta-sigma modulation. Most cheap players never made
it known what they used..and eventually when 1-bit was standard...they
stopped making a big deal (unless you had a specific trademark).
So the idea behind SACD was.."let's take the noise-shaping and conversion
process out of the player and feed it a delta-sigma modulated stream". The
rate increase was to squeeze a little more fidelity out of it..more
headroom for ultrasonic noise-shaping and an increase in dynamic range.
Original specs listed a response of just 30khz, this increased as
filtering got better.
They weren't going "we need to make some format with bat-fidelity
resolution"..it was more "we need to evolve the CD a bit to solve it's
current limitations." The details of exactly why a 64x over Redbook was
chosen are lost on me...and I can't be arsed to look them up. By the time
SACD came out...DVD was a thing and 96khz/24bit DACs had dropped in price.
But something else happened....they stopped using 1-bit delta-sigma in
standard DACs and started using multi-bit. So now things change...and your
conversion process from PCM gets that much easier and jumps up a bit in
quality.
Still....there is just *something* about a DSD stream that's different and
almost a tad "warmer" than a PCM stream. The general use DAC I have down
here in the listening studio I think uses multi-bit delta-sigma even with
1-bit input...but DSD still sounds a tad warmer. It's an ESS9018 based
chip...and I believe they have an updated version we need to check out.
Anyway...we always say "trust your ears, not your eyes."
So on to the second subject involving Paul Simon's high-res..no..we don't
have any specific details on it. But here's the.audiohut's opinion on what
you're wondering in general.
"If it's completely dead fucking silent up there...and I mean you filtered
out everything below 22khz and amplified the audio up to the point
quantization would be apparent...and there's nothing; then it's probably
an upsample of some sort. If they actually captured at a HD rate...then
you should see at least some noise. I mean..it's possible they could have
bandpassed the tape just for the traditional hearing range..it's possible
it could have been mixed like that. But there should be something; not
that it would indicate anything. They could have applied noise reduction
which made everything in the ultra-sonics quiet. They could have filtered
them out digitally if they wanted and claim it's still 192khz or whatever,
because technically it was digitized at that rate.
So the lack of audio up there might not indicate anything; it might
indicate an upsample. The problem is unless you go in and see something
like repeated samples..you won't know. And I haven't seen that shit
attempted since the mid 90s."
Yeah..it doesn't help much. Typically if everything stops pretty much
exactly at 22.050khz in the spectrum...it's an upsample. But like TAH
said...not everything was recorded to full bandwidth...and post-processing
could eliminate any analog noise leftover in the ultrasonics.
If you look at a 192khz release and see a band of noise at the top...that
is a DSD conversion though.
|
| Follow-ups: | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 |
|