Topic: Korg Mr-1 vs Sony PCM D-50

[SmokinJoe]

At the risk of oversimplifying and adding to the flamewar...

As I understand it, virtually all devices use a similar sigma-delta A/D converter type device/algorithm, so a D50, or MR1, or an R09 or a V3 all behave similarly up to a point.  The difference is what happens after that.  On PCM devices, it then takes this 1 bit output and runs it through their own variety of DSP to dither/resample to whatever format you chose in the menu (16/44 or 24/96, etc).  The Korg writes the raw DSD data to disk, and then you do the PCM conversion on your computer.  Whether the Korg software does a better job at home than the Sony/R09/V3 does in hardware/firmware is anyone's guess.

[Nick's Picks]

I never quite figured that out in the A to D chain.
the sigma -delta part is a 1bit stream that somehow makes it to 24 or 16 at whatever sample rates we choose on said hardware.
the whole point of DSD was a new approach to digital that cut out everything after the sigma/delta part and went with that output stream direct into its own formula.
either way,    I like them both.


from Wikki:
A Sigma-Delta ADC (also known as a Delta-Sigma ADC) oversamples the desired signal by a large factor and filters the desired signal band. Generally a smaller number of bits than required are converted using a Flash ADC after the Filter. The resulting signal, along with the error generated by the discrete levels of the Flash, is fed back and subtracted from the input to the filter. This negative feedback has the effect of noise shaping the error due to the Flash so that it does not appear in the desired signal frequencies. A digital filter (decimation filter) follows the ADC which reduces the sampling rate, filters off unwanted noise signal and increases the resolution of the output. (sigma-delta modulation, also called delta-sigma modulation)

The process of creating a DSD signal is conceptually similar to taking a 1-bit delta-sigma analog-to-digital (A/D) converter and removing the decimator, which converts the 1-bit bitstream into multibit PCM. Instead, the 1-bit signal is recorded directly and in theory only requires a lowpass filter to reconstruct the original analog waveform. In reality it is a little more complex, and the analogy is incomplete in that 1-bit sigma-delta converters are these days rather unusual, one reason being that a 1-bit signal cannot be dithered properly: most modern sigma-delta converters are multibit.
Because of the nature of sigma-delta converters, one cannot make a direct comparison between DSD and PCM. An approximation is possible, though, and would place DSD in some aspects comparable to a PCM format that has a bit depth of 20 bits and a sampling frequency of 96 kHz [2]. PCM sampled at 24 bits provides a (theoretical) additional 24 dB of dynamic range.

bla bla bla...., and they wrap it up with
"Sound quality of DSD vs. (24-bit/176.4 kHz) PCM

When comparing a DSD and PCM recording of the same origin, the same number of channels and similar bandwidth/SNR, some still think that there are differences. A study conducted at the Erik-Thienhaus Institute in Detmold, Germany, seems to contradict this, concluding that "hardly any of the subjects could make a reproducible distinction between the two encoding systems. Hence it may be concluded that no significant differences are audible". [4]"

[craig_c]

This is for those who have not yet seen it, here is a link to Korg's mini white-paper on DSD, what they call "Future Proof Recording".
http://www.korg.com/services/products/mr/Future_Proof_Recording_Explained.pdf

From the white paper:
There is no doubt that current 24-bit / 192 kHz audio sounds very good. But there are still areas that can be improved, and approached from a different perspective with other unique benefits.
The Multi-Bit PCM Encoding Process
• A/D & D/A conversion is done in low-bit (e.g.1-bit) at high sample rates, then one in X samples is kept (the decimation process)
• Decimation occurs during recording (i.e. A/D conversion) then interpolation and sigma-delta modulation during playback (D/A conversion)
Surprisingly, most current 24-bit converters actually use 1-bit conversion at the front end already. After capturing a high-speed 1-bit stream the converter uses what is called a Decimation Filter to change the 1-bit data into the desired multi-bit format.

It goes on to explain, in laymans terms, the difference between DSD and PCM.
The upshot is that, like some have presented previously, most PCM recorders DO capture 1-bit at the front end, and immediately process into a programmed PCM format. There is a nice little diagram showing the decimation filter process. Again, this is for us laymen, not you professional engineers who obviously already understand this process and its shortcomings. I'm not advocating DSD, just trying to elucidate for this dialog.

Craig

[headroom]

I asked my personal HIFI GURU about DSD

DSD drawback:
There is not one Programm that can fade in/out in DSD native, no EQ nada. Only PCIe Solutions with Hardware Software and expensive.

1 bit AD Proccesor
There is no such thing possible, it`s a marking lie. One bit or zero bits with a processor gives in the real world eighter 0% Distortion and no signal, or 100 % Distortion and 100% Signal.  1 bit Process is in reality a 4 bit sampling and therafter convertet in PCM or 1 Bit DSD. 1 bit gives maximal 6db  S/N.

In DSD  the Feeback loop of the DSD Signal has X-Grades for AD and DA.
The whole DSD Playback Noiseshaping Process is a constant up and down between Impulse and Noisesuppression  not a help for good sound.

DSD Is a lot more processed and complicated the PCM when you look closer in the real process. DSD is simpler then PCM this is only Marketing blaba.

[headroom]

Audio Engineering Society Convention Paper 5395

Why 1-Bit Sigma-Delta Conversion is Unsuitable for High-Quality Applications
by Stanley P. Lipshitz and Prof. John Vanderkooy Audio Research Group, University of Waterloo

Top Audio Engineers worldwide brings Ultimate Nails for the Coffin DSD/ SACD

http://sjeng.org/ftp/SACD.pdf

Professional Musicians with trained ears have swichted to SONY PCM 50 after the listening expirience with KORG recorders

[Shadow_7]

May of 2001 really?  My device is dated 2006-ish and does 2x's the sampling rate that paper is based on.  Most of thier distortions being above the kHz that's even reproduceable on conventional speakers.  Is it theoretical perfect, I guess not.  Is it better (at 1GB per 12+ minutes, versus 1GB per 1+ hours for CDs), it is IMO.

I suppose that if you're feeding a DSD recorder with the line output of a CD player, no it's not.  Since the source material is effectively the same as a CD recorder.  And a lot of microphone preamps limit their Hz ranges to not make use of the frequencies that could be sampled at DSD bitrates.  But I like my DSD it's clearly better sounding IMO.

I suppose that if the paper was less biased, I might have read it more thoroughly.  But PCM is infinitely perfectible and the opposite argument for DSD.  Clearly biased, i.e. someone with an agenda.  DSD is just another file format IMO.  Maybe there's a lot of junk in that bitrate, maybe there's not.  Once the editing software catches up I guess we'll actually know and not deal in theories.  And audiogate has made some pretty good strides recently.  (if only they had a linux version)

[H₂O]

I also don't want to jump into this "bon fire" of swinging fists over different ways to skin a cat, but personally I ran my Korg MR-1 against my Sound Devices 702 and prefered the sound of the MR-1 - ended up selling my 702 -> I have external Pre's that I prefer much over the internal 7XX pres so the ALL-IN-ONE factor was not a deal with me.  Compared - 24/96 on the 702 and 1 bit converted to 24/96 on the MR-1.
 
FWIW - 7XX's use a more advanced (i.e. a multibit Delta/Sigma) A/D versus standard 1-bit Delta/Sigma A/D such as the Burr Brown models.
 
 

[craig_c]

I'm not sure why this has turned into a debate over whether DSD is a worth a s**t, but clearly at the high end of recording a lot of engineers and studios are using it.
Reference:
Tasacam DV-RA1000HD
EMM Labs
Sonic Studio
Phillips
Sony
Meitner
Sonoma


Craig

[headroom]

For shure DSD Recordings are good but not High End even when the big shots are involdeved
The blind test result speakes for themseself DSD Recordings versus CD most people prefer the CD, someting is wrong. People cannot hear it, because of low Audio chain resolution and/or it is DSD system inherent bot oossible to hear it.

Audio Asylum Thread Printer Hi-Rez Highway SACD just a few comments

A common conception about SACD is that a 1 bit system needs to sample at a rate sixteen times faster than a 16 bit system. However, the SACD 1 bit system is not a weighted system. For example, common rs-232 transmission is via 1 bit, but they are arranged in groups of 8 that make up a byte. Each bit is weighted as a power of 2.
The SACD bit can be thought of as an input to a counter. If the bit is 1 you count up. If the bit is 0, you count down. Therefore, it would take 256 1's to count up to 256, not 8 1's. To count to a 16 bit number takes 65535 1's.
So, in short, the SACD sample rate must be 65536 times the 16 bit sample rate. The equivalent of 44.1 KHZ at 16 bits is 2.8 Giga Hertz at 1 bit.
Or to look at the inverse, at SACD's 2.8 Mega Hertz rate you can describe a 20 KHz signal with 2,800/20 or 140 levels. This is less than the range of 8 bits.
This is not to say that the 1 bit approach doesn't have better low level resolution.

The s/n ratio from scad players above 12kHz is generally lower than that of a 16bit cd player due to the rise in noise floor from 8000Hz and higher.
The signal change rate coming from the DA converion has a finite resolution so sometimes the resulting amplitude is a bit to high and after another 5 or 7 bits it's a bit to low and so on and so on. 
This is the way how the original signal wave is closely reconstructed. 
The little over and undershooting happening during the wave reconstruction is the cause for the 'dsd' noise.
This is all a very cunning balancing act to keep the dsd noise just below the human hearing threshold.

[it-goes-to-eleven]


The SACD bit can be thought of as an input to a counter. If the bit is 1 you count up. If the bit is 0, you count down. Therefore, it would take 256 1's to count up to 256, not 8 1's. To count to a 16 bit number takes 65535 1's.
So, in short, the SACD sample rate must be 65536 times the 16 bit sample rate.

Interesting.  I wondered about that, and was thinking about how it compared to pcm encoding.

I think there is a lot of hype around dsd (and, in the past, marketing dollars).  But the lack of playback options, and the tight control of the format, keep it from getting fairly evaluated.  Especially with unprocessed live material.

[ghellquist]

And again headroom throws a lot of theory. I cannot emphasize enough that a little knowledge is very dangerous. And in these cases each and every theory has to be subject to the real world tests. There is also a problem that headroom draws conclusions that not really can be drawn from the theory I will not go into a pissing contest here, but I suggest people to read his posts with a good sprinkling of salt.

One user, h20, has reported liking the MR1 in tests against a 7xx. This is a valid test based on personal experience. I have done a similar test with my 722 against a MR1000 running both internal and external preamps - I continue to use the 722 based on personal taste and convenience of everything in one package.

// gunnar

[headroom]

Yes yes all theory and snafu

http://www.stereophile.com/thefifthelement/1104fifth/index2.html

PCM vs DSA aka SACD
Again, any processing of the 1-bit stream on the user's end assumes that it will be available at the output of the SACD player. For copyright reasons, this is something that might never be made available.

What is clear is that the Sony/Philips Super Audio CD system does not perform to the same level as the best LPCM converters now available. The best ADCs' delta-sigma modulators, for example, produce four times the data per second as DSD, which means a lower noise floor and lower levels of distortion than can be produced with DSD (assuming the same loop-filter order)

[headroom]

Brainfood for non biased people

http://sound.westhost.com/cd-sacd-dvda.htm

[it-goes-to-eleven]

Yes yes all theory and snafu

http://www.stereophile.com/thefifthelement/1104fifth/index2.html

But what's your point there?  That article just documents that a particular SACD release is apparently the 16 bit pcm audio run through a DSD converter..  And while completely bogus, that says more about the company than about DSD.  Though it does appear to be a case of trying to cash in on dsd hype.

And if the original master is PCM, I don't see the point of converting it to dsd.

[Shadow_7]

Numbers games.  I used to be good at those.

16 bit * 48000 Hz * 60 seconds * 60 minutes == 2,764,800,000 bits per hour
/8 (bits to bytes) == 345,600,000 bytes per hour ~= 350MB per hour
*2 (mono to stereo) ~= 700MB per hour (sounds about right)

Assuming Hz as a per second multiple.  Not all that up on the techno babble.

((24 bit * 96000 Hz * 60 seconds * 60 minutes) / * 2 ~= 2GB per hour

-----

Going backwards based on resulting files.

12 minutes 41 seconds == 1,074,266,204 bytes (DSD @ 5.6MHz)
according to audiogate and windows explorer respectively.
DSD is supposed to have more dynamic range than 24 bit PCM.  So lets call it 32 bit (or not).

12*60 + 41 == 761 seconds

1,074,266,204 bytes / 761 seconds == 1,411,650 (and some change) bytes per second.

1,411,650 * 8 == 11,293,200 bits per second

11,293,200 / 2 == 5,646,600 for mono (5.6MHz imagine that)

5,646,600 / 32 bit == 176,456 (and some change).  Assuming 32 bit, still > 96kHz by the looks of it.

5,646,600 / 24 bit == 235,275 (even).  ~= 2.4 times the rate of 96kHz.  still greater than 192kHz.  But there are PCM devices that do 384kHz now.  And I'm not entirely sure of the bits spec for DSD when converted.  24 bit has greater dynamic range than 16 bit.  DSD according to the Korg manual has greater dynamic range than 24 bit, so anybodies guess.

Stiill 200kHz is better than 96kHz?  At least in terms of bits.  Having a mic and source that uses that bandwidth?  Does a PA system really use it?  Versus something more acoustic like a stringed instrument or brass section?  Does it matter if your mic or preamp doesn't send bits > 48kHz down the line?  Can we expect a new line of HD microphones in our future?  0Hz to 192kHz versus 20Hz - 20kHz?  Who knows a lot of it IS just theory.  In terms of ADC / DAC, my Korg is the best I've got right now.  It's converters are tops IMO.  Even if it's not dealing in DSD.