Lame. Skepticism is boring.

Yeah because being fooled by unscrupulous marketeers is the height of excitement.

Alistair

I dare anyone here with highend dac and amps to get a devialet 200 on demo.

Listen to it. Vinyl, analogue, digital. Anything.
It sounds a lot more transparant than what you and I currently have.
Do not believe me. Just try it.

And the fun thing is, standard setting has it ADDA everything internally at 96K...

Of course you can connect an analogue output of a dxd recording to it.
It will still sound essentially transparant.

I was utterly impressed.
The difference has NOTHING to do with what some hear between 44K or 196K.
The difference between my current amp (no slouch either) and the devialet amp is just mind boggingly great.

Something in the order of half a century.
Almost the entire audiophool world press even agrees its internal MC input (which is by definition digitalised !) sounds great.
At 96K.

For Undertow: Nonsense! You don't understand the subtleties of the English language. The very first thing that Grimm paper says is a release at native 8fs speed is an outstanding idea. Read the paper again. The Grimm paper is a good reminder now because of the tsunami of dsd myths propagated by the hi-end hi-fi press and manufacturers. HOwever, Grimm are not MY gurus.

I base what I said about DXD on the playbacks I heard during the Hi- Res Forum at the AES show in Los Angeles last month. Did you hear it?

What I wrote is my informed opinion from what I heard recently on the coast. AND from what I hear in my listening room with Merging equipment (Hapi) + PMC monitors set up.

{edit for clarity}--to clarify, yes, I can work with 384kHz. rates and dsd in my studio===> I do it on Pyramix with Hapi.

Comparing the DXD and 5.6 MHz DSD files of some 2L recordings, I find that the DXD sounds slightly more transparent. But DSD-capable converters are still rather more common than DXD-capable converters. I have not compared the DXD recordings to 4fs or 2fs files, but I'm wondering whether DSD or 4fs PCM would be the preferred consumer-friendly high resolution format for those who do not have a DXD converter. (I do, for what it's worth.) The Grimm paper suggests that PCM might be the better choice, but 2L issues their recordings in just about every file format from DXD and 5.6 MHz DSD right on down to MP3.

How can you judge the transparency of something without having an original?

I am a native-born English speaker. I have done a lot of reading in this language. I had to, my graduate degree required me to read at least one book, usually two and sometimes three a day. I can honestly say that I have a handle on the language and its nuances. However, the DSD paper is not subtle or nuanced. In its conclusion it says::

"A final remark. If in our opinion 192/24 files
are a better choice than DSD files, then why do
we include the ‘DoP’ DSD format in the USB
interface for our LS1? The reason is simple.
Until recently there were just two important
formats for musical content, CD’s with 44.1/16
and SACD’s with 64 fs DSD. A lot of wonderful
music has been released in these formats and
recently the SACD masters are becoming
available as online downloads. We want to
enable our customers to enjoy that music and
therefore we will support every quality format.
In our view it is music over format, not the
other way around."


BTW, I wish you would stop belittling people as a gambit in your attempt to triumph in an argument. You must be smart enough to not have to resort to ad hominum attacks.

Fair question. I should clarify that I heard a bit of coloration in the DSD signal, something I have experienced in some other DSD recordings. It's very subtle, and (as the Grimm paper notes) some people find it euphonic. The DSD recordings sound fantastic, but the DXD sounds even better. However, we are definitely splitting hairs at this point.

Plush is more than capable of standing up for himself, but I think his argument was not in favor of DSD. He was championing _DXD_, which is PCM at 8 * standard sampling rate.

The argument I get from the Grimm paper is that ideally, one should avoid unnecessary conversions between PCM and DXD, and that since most DSD recordings involve a conversion to PCM at some point, DSD does not have advantages as a release format. They do maintain that 4fs or 8fs PCM does not have advantages over 2fs as a release format for DSD-sourced recordings, but I do not seem to recall them saying anything about disadvantages of native 8fs PCM recordings.

Note that they do not disparage DSD when used as an archival format or if the recording is mixed in the analog domain and the editing is minimal. (Several other labels, besides Channel Classics, have taken this approach. Telarc comes to mind.)

I am curious about their dismissal of DSD128 and DSD256, however. It was my understanding that the purposes of these formats is to move the ultrasonic noise inherent in DSD further out of the audible band. While this theoretically might not be audible, it should remove the need for sharp filters close to the audible band, and that might be audible under at least some circumstances.

Fie upon your argument and petty scolding!

Sorry, but you too are wrong. I am talking about DXD which is pcm, not a variant of dsd. The paragraph you cite is not applicable to what I argue.

I am a rhetorician so I am, more than the usual person, focused on accuracy of content when summarizing written communication. That is why I have little tolerance for summaries that inaccurately portray the meaning and conclusions of what was written. (in the Grimm "white paper")

If an attack on what I have written inaccurately summarizes the source material, you bet I will call it to the attention of the writer. It is very clear to me that there was an incomplete understanding of the English language by some posters. Then they went on to offer a refutation of what I wrote based on their mis-reading. Not the worst thing in the world but not something to hang your hat on when trying to refute something.

It's also true that there would be a gross incomplete understanding on my part if I were to try to understand some of the native languages spoken here by posters.

Nice try but I am British. (My name should be a hint to anyone but the least observant of native English speakers). I also speak Dutch at native level so I am likely to catch any Dutchisisms in the paper.

You on the other hand seem to have trouble with the absolute basics of the English language. (See below).

Not only is the very first mention of 8fs or DXD nowhere near the first thing mentioned in the paper, unless you consider half way through page three of a four page paper the beginning, but the very first mention of it states the following:

There is no subtlety or ambiguity in the phrase "4 fs and probably even 2 fs can be transparent as well". How do you translate that paragraph to "8fs speed is an outstanding idea"? To me that translates loosely to "8fs is overkill". Other comments in the paper make it clear that Grimm believe anything above 4fs to be overkill.

Why don't you start by reading it once instead of skimming it for titbits that confirm your pre-existing world view.

The Grimm paper is a good reminder that DSD is a marketing format and is inferior to PCM. The paper is rather clear about that even though it remains diplomatic about the subject.

I don't have any gurus. I don't need them. I can actually think for myself and I understand how digital audio works. That said, I do not pretend to know everything and I am always open to new perspectives and evidence so, when certain people speak or write, I pay attention because they have proven over time to be intelligent, knowledgeable and intellectually honest. Grimm Audio fall in that category.

I've heard 8fs, yes, and i've heard it in a much better environment than a presentation set up by the manufacturer of a product[1] at an AES show. ([1] Just a guess).

Wait a minute, you are comparing what you are used to hearing in your own studio with something you heard at an AES show or you have 8fs in your studio?

To be clear, I am not denying anyone has heard great DSD and/or 8fs converters. What I am suggesting is that the actual sample rate and/or format have very little if anything at all to do with the quality they heard.

Great analogue sections, great clocking, great, power supplies, smart layouts, quality converter chips and/or components, extra powerful DSP chips to take care of filtering and any other processing involved, high quality DSP algorithms and code etc. Those are the things that count when considering converter designs and implementation. Sample rates and formats are just marketing for the technically naive.

Also, I am not talking about personal preference. Some people might prefer the sound of some DSD implementations (or 8fs converters) for whatever reason. I am talking purely from the perspective of transparency and accuracy.

The one advantage of DSD is that it removes some steps in the converters. That is an advantage to the _least_ competent designers because it means there are less things that they can mess up. Competent converter designers are capable of creating transparent converters at 2fs and, based on all the evidence available, even at 1fs.

Alistair

Sorry to be a pain but how do you come to the conclusion that what you hear in the DSD signal is a coloration instead of something that just is there and a correct part of the signal?

Again, a fair question. If there is a quality present in the DSD recording that is not present in the (original) DXD recording, than it is fair to assume that it is a coloration or artifact of some sort. In this case, the DXD is the original recording, and the DSD file was sourced from the DXD.
Note that 2L themselves state that they feel that DSD is not completely transparent (their word): "We find DSD, as used in the SACD format, somewhat different in colour from PCM; in some mysterious way DSD is softer and more beautiful but slightly less detailed....I personally prefer extremely high resolution PCM over the DSD and I would claim that DSD is not transparent." (Source: http://2l.no)

Note that they also claim that the differences between 96, 192, and DXD are subtle.

Not sure if I have dived into this subject before but ....

Let me make to following observation. I am not aware of the conversion processes implemented in DXD hardware. However, I am not aware of a native AD pure PCM convertor at 384kHz (except possibly in the bowels of some secret military skunkworks). It may be safe to assume therefore that the A-D conversion used for DXD is still of the Delta Sigma variety. Thus it uses right at the start of the process, the same conversion technique incorporated in DSD.

Direct PCM conversion (when acheivable) actually coverts the amplitude of the signal at that point in time, with no influence from previous signal history. Its overload characteristic is peak instantaneous amplitude. Adaptive delta modulation (DSD or Delta-Sigma) relies on previous signal history (through integration) to produce the necessary variable step size to accomodate the required dynamic range without sampling at satellite communication frequencies. Its limitation (in the first instance) is slope-overload. Early research shows the ear is more tolerant of slope overload than peak clipping.

Perhaps the makers of DXD instrumentation hardware can advise if their conversion process is pure PCM, or the more common hybrid of Delta Sigma. It is worthwhile to know what we are listening to before debating merits.

And, in the end* what we are listening to is music (with possible exception of the creators of 'breaking glass' movie sound tracks ... )


[*the love you take, is equal to the love you make.]

While a number of Grimm articles provide interesting technical and/or historical context for the reader, one should never forget the common thread of these papers (at least the examples to my knowledge).

Inevitably there is the subtext of positioning it's product against others. Another example: http://www.grimmaudio.com/site/asset...nds_or_ppm.pdf

No argument there are useful facts to be gleaned from their articles. However, without diving deeply into the technical aspects (keep in mind the intended audience), and providing certain assumptions and narrowing the examples, the reader is left wanting for a more transparent discussion.

While Grimm may well be motivated to dispel audiophile myths, is it not in the end further contributing to such, considering it too must be considered in this audiophile class?

(In all fairness, this contribution might not be intentional, but rather understood as misunderstanding that exists without exhausting the subject. Factor in an eye catching headline... ).

Arda Technologies - AT1201 Flagship Audio A/D Converter

From the ARDA 1201 ADC specification:

Note that the only way to operate at octal speed mode [8x fs] is for the MCLK frequency to be 512x, or nominally 24.576MHz.

For linear (non-integrating) delta modulation to deliver 20 bit resolution within the sampling interval corresponding to 96K, it would need to sample at 100,663,296kHz. Integration (varying the step size guided by signal history) reduces the required sampling rate (but introduces the opportunity for non-linearity in the integration process). Consideration of spectral energy distribution may also allow a lower sample rate.

It is possible to keep developing more and more extravagent conversion processes, and the market will decide. But perhaps we need to revisit the history of Apogee and perhaps pay closer attention to anti-aliasing/decimation and reconstruction filter techniques.

As the block diagram shows, it is indeed a Delta-Sigma front end. However, it is multi-bit which avoids some of the shortcomings (Lip****z et al) of single bit conversion, which is the basis of DSD.

How insightful to bring out the multi bit delta output, allowing others to investigate new approaches to integration and filtering. Maybe someone could do a Sony (a la DSD) and propose a distribution standard based on the unprocessed multi-bit signal for the purists or researchers. Just needs a 73.368Mbit/s bandwidth. Can we fit Beethoven's Ninth on a single Blu-Ray?

Resurrecting the thread, I know.
Mr. Seeto,
I wonder how they implement post ADS fine gain adjustment when recording to DSD?

I believe it is all done within the ADC chip. The link is the data sheet for the product which is an obsolete product.


https://www.profusionplc.com/images/.../at1201-ds.pdf

Best,

Da-Hong

SACD is only DSD64, I beg to differ. DSD128 on the Korg MR-1000 sounds transparent compared to analogue source, in the headphones (Senns HD580).

DSD people will say: measure, think and design all you want, we will be glad when you get back with any digital format that is transparent to the source.