Post by RWetmore August 9, 2008 (21 of 43)
|
|
tailspn said:
Could you please provide a credible reference that states that? That would be a great surprise to the various companies that actually design and manufacture DSD A/D converters.
Tom
Wikipedia isn't a credible reference? That is at least how I read it and how I've heard it described on various forums. I could be mistaken though since I haven't actually worked with it. Either way it's highly unlikely we would ever get such a format anyway.
|
|
|
Post by RWetmore August 9, 2008 (22 of 43)
|
|
I have long standing theory that the key to a truly transparent digital audio system is one that requires no filtering. 441khz/24bit would do that from what I understand.
Many of the problems inherent in DSD and PCM are well articulated by Peter Scheelke (from DAD) in this very detailed post of his here:
http://www.gearslutz.com/board/1712442-post30.html
|
|
|
|
|
|
RWetmore said:
I have long standing theory that the key to a truly transparent digital audio system is one that requires no filtering. 441khz/24bit would do that from what I understand.
Many of the problems inherent in DSD and PCM are well articulated by Peter Scheelke (from DAD) in this very detailed post of his here:
http://www.gearslutz.com/board/1712442-post30.html
Unless I’m mistaken, most A to D converters use the Delta Sigma design. And after studying how these converters work it appears to me that the very first stage is a bit stream. It is then up to the designer to represent this bit stream in the coding format they choose. SACD simply records this bit stream for the most part directly to the storage format. Thus eliminating the need for a PCM or DXD (same thing) conversion. As I mentioned before, editing makes it necessary for this conversion. I’ve read that there is a one bit editing process but I have not studied how this process works. I suspect that the DXD converter that you referenced does this PCM or DXD conversion on board instead of inside a workstation. Could you please provide a link to the DXD converter circuit design?
|
|
|
Post by tailspn August 10, 2008 (24 of 43)
|
|
RWetmore said:
I have long standing theory that the key to a truly transparent digital audio system is one that requires no filtering. 441khz/24bit would do that from what I understand.
Many of the problems inherent in DSD and PCM are well articulated by Peter Scheelke (from DAD) in this very detailed post of his here:
No, I would hardly consider Wikipida a credible reference for such a narrow highly technical small community subject such as DSD, 1 bit modulators, and dithering technologies. Unless of course, you're interested in the man on the street view. A much more realistic insight can be gained from published papers from recognized professional societies that at least have some amount of peer review. Mail me and I'll send you a list of the seventeen papers I have, including the Lipshitz series. I'd highly recommend the Reefman Modulator Tutorial AES paper that followed the Lipshitz.
|
|
|
Post by tailspn August 10, 2008 (25 of 43)
|
|
Your Peter Scheelke reference is a bit self serving to be considered completely factual. Peter is in the business of promoting his DAD converters which includes his proprietary DXD (PCM) format in addition to DSD and various traditional PCM. He is not correct in stating that "DSD cannot be edited since it is only 1 bit." The Sonoma DAW, which Sony seeded the market with more than a hundred systems, edit completely in 1 bit using the Sony developed E Chip. Since Peter's major competition is the installed base of those Sonoma DAW systems, I find it disingenuous that he makes that statement. He is factual when he states that the impulse response of 64fs DSD is "as close to perfect as we have ever seen," and his DXD is almost as good "(88%)." Myself, I go for the impluse response, especially since Peter has shown that with his DAD converters converting DSD that he's "been able to keep the noise down until 23 kHz (less than -120 dBfs measured 20-20 kHz), however the energy from the noise is still part of the signal, so after 23 kHz the noise becomes increasingly significant." As the Reefman paper shows however, the noise is still more than 50dB down at 100kHz, but could be a problem with some equipment, notably some power amplifiers. IMO, if you really seek perfection, purchase good components rather than cripple the signal by choosing less than optimum formats.
|
|
|
Post by tailspn August 10, 2008 (26 of 43)
|
|
But I'm picking on Peter. He has an extremely valid practical point. Editing pop music in DSD is problematic, and pop music is by far the largest share of the recording industry. While you can do all the editing in 1 bit DSD necessary for classical and Jazz, the Sonoma and SADiE DAW editing systems, that use the E Chip, still don't look like Pro Tools, and can not do EQ and many many of the special effects that are the core of pop music. Also, the multi layering techniques used in studio pop recordings put DSD at a disadvantage due to noise buildup with each generation of layering. Perhaps Peter's PCM DXD is a great alternative to DSD for the major pop recording business, and he'll make a fortune. For me and my interest in acoustical music, 64fs DSD is the closest to the original signal delivery format available, or on the horizon. 128fs DSD would be theoretically better, but probably only for those few suffering power amps.
|
|
|
Post by tailspn August 10, 2008 (27 of 43)
|
|
But there is another issue that is at the core of your recent postings. The thought/hope that a high rate (441kHz?) 24bit PCM would cure the ills you believe exist in the available formats. On the surface, that's a valid point. 882kHz 32bit or higher would be even better! Right? Also I believe what you're talking about here is a delivery system, not a all-in-one box editing system format to be later converted into a delivery media like 64fs DSD, as is the practice with DXD. Putting aside the problems inherent in a PCM format, and understanding that this is a consumer system, not a cost plus government technology project, we encounter the well documented problem of jitter with increasing data rates. In a DAW editing station, format conversion/data handling of say 500KHz 24bit data can probably be handled with acceptable jitter, especially if it's all done on one card with one clock source. It's simply the product of bits times time. Six channles at 500K 24bit words is 72Mbits/sec. That's 13ns per bit with six channles. There is however no consumer delivery system technology present, or on the horizon, that can deliver that data rate with acceptably low jitter through a distributed transmission system like optical storage. There is a growing consensus in the recording biz that believes there is a PCM sampling rate sweet spot around 100KHz that's optimum sounding. Faster than that, and the quality begins to degrade. Jitter, plain and simple.
And THAT's the beauty of SA-CD. A relatively low data rate, (2.8Mbits/sec) with a high sampling rate (2.8MHz) with subsequent gentle filtering. Perfect impulse response (the real thing that counts), and it's here now. And it's why you have posted numerous times that you prefer the sound of your SA-CD's to other formats you've heard.
Tom
|
|
|
Post by RWetmore August 10, 2008 (28 of 43)
|
|
tailspn said:
But there is another issue that is at the core of your recent postings. The thought/hope that a high rate (441kHz?) 24bit PCM would cure the ills you believe exist in the available formats. On the surface, that's a valid point. 882kHz 32bit or higher would be even better! Right?
No I don't think it would be any better, and there would definitely be no improvement going from 24 bits to 32 bits since even the best A/D converts can only do about 20 or 21 bits of resolution. The remaining 3 or 4 bits are just recorded as noise. Even further, most consumer gear can only do about 15 or 16 bits of resolution.
The idea behind 441khz is that it's the first multiple of 44.1khz that supplies a frequency response high enough where no filtering would be required. With such a sampling rate, one could record and play back without any processing whatsoever. The hope is the lack of any processing would eliminate any digital "glare" or distortion that such things cause. The data could also be directly down-converted without any distortion to 44.1khz/16bit for release on regular CD.
|
|
|
Post by RWetmore August 10, 2008 (29 of 43)
|
|
tailspn said:
Also I believe what you're talking about here is a delivery system, not a all-in-one box editing system format to be later converted into a delivery media like 64fs DSD, as is the practice with DXD. Putting aside the problems inherent in a PCM format, and understanding that this is a consumer system, not a cost plus government technology project, we encounter the well documented problem of jitter with increasing data rates. In a DAW editing station, format conversion/data handling of say 500KHz 24bit data can probably be handled with acceptable jitter, especially if it's all done on one card with one clock source. It's simply the product of bits times time. 500K 24bit words is 12Mbits/sec. That's 83ns per bit. There is however no consumer delivery system technology present, or on the horizon, that can deliver that data rate with acceptably low jitter through a distributed transmission system like optical storage.
What about Blu-ray disc? It should be able to handle it with ease and with a ton of bandwidth to spare.
|
|
|
Post by RWetmore August 10, 2008 (30 of 43)
|
|
tailspn said:
There is a growing consensus in the recording biz that believes there is a PCM sampling rate sweet spot around 100KHz that's optimum sounding. Faster than that, and the quality begins to degrade. Jitter, plain and simple.
And THAT's the beauty of SA-CD. A relatively low data rate, (2.8Mbits/sec) with a high sampling rate (2.8MHz) with subsequent gentle filtering. Perfect impulse response (the real thing that counts), and it's here now. And it's why you have posted numerous times that you prefer the sound of your SA-CD's to other formats you've heard.
Tom
Yes, but a 1 bit system is fundamentally flawed and can never be perfected as documented by Lipshitz, Vanderkooy, etc. It sounds good yes - better than CD when done well, but it's not clear this "better" sound is achieved through improved transparency to the source. Is the perceived better sound primarily do to us liking how the 1 bit distortion sounds or because of the superior impulse response and frequency response? From what I understand, the majority of the scientific community believes it is the former.
Even so, it's not a good permanent solution if for no other reason than its noise shaping characteristics, which are gimmicky and unnatural.
But you are right, it is here now and can sound great. It's the best we've got at the present time.
|
|
