|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
65.19.76.104
In Reply to: RE: For at least the 100th time, NO posted by ahendler on January 28, 2015 at 07:48:41
The Mytek Stereo192-DSD does not sound good when playing 44/16 music. It sounds better when HQPlayer upsamples 44/16 to 88/32, better still when upsampled to 176/32, and better still when upsampled to DSD128. (Upsampled to DSD64 sounds slightly softer than upsampled to 88/32 or 176/32, something that is not quite right. DSD64 or DSD128 playback of pure DSD recordings produces the best sound of all.
Another aspect of the Mytek is that it sounds best when the analog volume control is bypassed. I send its output to my amplifiers (powered monitors) and so I need to use a volume control. I use the digital volume mode that the Mytek provides. The Mytek has many options and features and unfortunately it takes some effort to figure out which ones produce the best sound. For example, I found it necessary to open up the case and add the jumpers so as to reduce the output level by 6 dB.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Follow Ups:
Hi there, in a long diatribe against SACD, on www.iar-80.com, Moncrieff pretty much slams the DSD model, saying the sound is more homogenized and less informative in the treble, and saying the amount of information is actually less than in the PCM system; DSD is a one bit system that mangles the signal to come close to 20 bit resolution.
I'm not an engineer, but I've never been overly impressed overall with SACD sound, there's less 'there' there, it's phasey sounding, and it's not as immediate and punchy as PCM.
His main argument seems to be based on equating the number of samples per half cycle with bit depth, which is a laymans misunderstanding of sampling. From that basis, he concludes that DSD has 6-bit equivalent resolution at 20 KHz, 7-bit at 10 KHz, etc. down to 12-bit at 312 Hz. This conclusion is nonsense. The equivalent resolution is a function of the noise shaping curve, which was optimized to theoretically provide better than 24-bit PCM performance up to 20 KHz, with a huge amount of ultrasonic noise above that as a consequence.
He also misunderstands noise shaping and confuses it with signal averaging, which leads him to claim that it destroys transients via averaging them out. This is also nonsense. Noise shaping filters the quantization error, not the signal. The transient response of a system using noise shaping will be commensurate with it's bandwidth, same as without noise shaping. DSD produces impulse responses and square waves like you would expect from its bandwidth. They look similar to what you get with 24/192 PCM, except at low levels they are fuzzed up with ultrasonic noise.
I would like to believe your analysis is correct. However, you should look at some of Thorsten's posts. This is a complex subject and "simplistic" analysis can sometimes be more useful than more sophisticated analysis, especially if the sophisticated analysis is made by people with skin in the game. This is not a simple issue. There is little published literature on the theory of DSD and those that exists say that the 1 bit format is flawed because it can not be properly dithered. DSD can work well, but it depends on a lot of "secret sauce" algorithms.
I tried to prove Thorston wrong. I failed. I was unable to convert a 44/16 Khz PCM signal to DSD and back to PCM and get close to 16 bits of accuracy, let alone the 20 bits claimed by Philips. I considered the peak errors, not a bunch of "averages". I was unable to get 16 bit accuracy even starting out with a DC signal. (I did this to avoid filtering issues.) I used a published DSD modulator. What I am saying is that the Philips information was marketing spin, not technical reality.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
To the extent that Moncrief uses theory to show that DSD64 is not as good in the treble, I think he is off base. Furthermore, I don't think it is possible to understand what is going on without a lot of heavy math, which explains how dither works. Moncrief doesn't get into this, nor would this be possible in a popular newsletter.
In practice, I put DSD64 roughly at 176/24 in terms of subjective sound quality, i.e. way above 44/16. However, this applies only to pure DSD recordings, not ones made by upsampling PCM, of which there are many examples. I try to listen to formats identical with the masters to the extent possible, because conversions to/from PCM and DSD are generally a poor idea. (Comment applies to DSD64, not to DSD128 or higher.)
Incidentally, Moncrief was my audio buddy in high school. We used to lug our tape decks, preamps, amps and KLH-6 speakers around the campus of our school so we could listen to symphonic music in large rooms. We were both introduced to Hi-Fi by none other than E. Bradley Meyer.
Note: DSD was a Sony/Philips marketing buzzword. The reality is that DSD64 is nothing but 2822.4/1 PCM. As such it benefits from a high sample rate, requiring no anti-alias filters on record and minimal filtering on playback. However, aggressive noise shaping is required and there is a lot of "secret sauce" to get good sound out of the 1 bit format.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: