Mobile Fidelity reissuing Michael Jackson's "Thriller" on SACD

QuadraphonicQuad Home Audio Forum

Help Support QuadraphonicQuad Home Audio Forum:

marcb

600 Club - QQ All-Star
Joined
Nov 15, 2010
Messages
664
Tomato tomato, the point is they don't do them.
Apples, oranges. I would say the implication was why they don't do them. And it's not because they are "anti-multichannel".

If they could do Mch at no extra cost, I'll bet they would.
 
Last edited:

marcb

600 Club - QQ All-Star
Joined
Nov 15, 2010
Messages
664
No one said it did.



It's not.



It's doubtful anyone can.
How do you know it's not an artifact?

Honestly, how (or why) did anyone even find them? They're just a few bits. I guess something like click repair would pick them up, but why would anyone run CR on a digital file?
 
Last edited:

dabl

400 Club - QQ All-Star
Since 2002/2003
Joined
Sep 1, 2003
Messages
423
Very interesting about the SACD peaks on some tracks.

I have a rip of the SACD and hear very audible problems on some of the kick and snare hits at the beginning of 'Billy Jean' using foobar with foo_dsd_asio and a Topping DSD DAC. It sounds like a popping sound and/or a bit of phase cancellation. It's not subtle at all and I figured maybe a better DAC might handle them better or possibly there's some settings I could change somewhere. This is the only track I've noticed problems on out of many dozens of SACD rips.

I've always wanted to test a 'print' of the sacd-r but unfortunately my Sony UPB-X800 won't play those

Mod Note:
Please Do Not share copyright material.
 

Cheezmo

800 Club - QQ All-Star
Joined
Jun 5, 2011
Messages
886
Location
Texas
I'm not sure what that has to do with how good the original SACD sounds, but...ok. Regardless it's probably just an artifact of converting from DSD to PCM. And can you hear them? I can't.

An artifact unique to that one particular SACD? No, it was authored with glitches and I am in complete agreement that you are unlikely to be able to hear them. I know I can't after I smoothed them out with Audacity ;-)
 

Cheezmo

800 Club - QQ All-Star
Joined
Jun 5, 2011
Messages
886
Location
Texas
How do you know it's not an artifact?

Honestly, how (or why) did anyone even find them? They're just a few bits. I guess something like click repair would pick them up, but why would anyone run CR on a digital file?

I found them because I noticed that when they were normalized, a couple of tracks had extremely abnormal peaks. It showed up in my normal work flow as a strange outlier. I looked the waveforms, found the weird glitches. Fixed them then went on with my life.

Doesn't change the fact that the glitches are there.
 

ssully

2K Club - QQ Super Nova
Since 2002/2003
Joined
Jul 2, 2003
Messages
3,380
Location
in your face
How do you know it's not an artifact?

Of dozens of rips done on SACDs released over two decades of time, this is the only one to ever display such an unfixable 'artifact' of ripping to me. So it's not an artifact of ripping. Something is inherently different about those tracks on the disc itself.

Honestly, how (or why) did anyone even find them? They're just a few bits. I guess something like click repair would pick them up, but why would anyone run CR on a digital file?

It has to do with 'best practices' for SACD mastering.
DSD 0 dB is specified as 50% modulation index that translates to -6 dB PCM level in direct conversion. And short term peaks are specified to be max +3 dB which is 75% modulation index and translates to -3 dB PCM level in direct conversion.

The foobar2k SACD plugin allows you to monitor conversion from DSD to PCM. The plugin lets you specify DSD vs PCM peak level before conversion -- you can use a range from +0dB (no change in peak level) to +6dB boost in the PCM, in 1 dB increments. You could, of course, simply rip at +0 dB -- peaks of the PCM version will then be between -6dB and -3dB PCM. But you may prefer to exploit the available PCM level range. That means you may must increase the level of the conversion by as much as +6 dB, so -6dB peaks are now at PCM 0 dB. But if the SACD mastering has any short term peaks at +3dB DSD (as allowed in SACD spec), and you specify +6dB of boost for PCM, you will get 'overs' in the PCM file by as much as 3dB. This will be an 'artifact of ripping'. Luckily the conversion can be monitored in foobar , using its Console. Overs will be flagged. That tells you that you raised the conversion level too high: try a smaller value, until the DSD peaks no longer exceed 0dBFS PCM. You are then utilizing the maximum 'dynamic range' offered by PCM.

The thing is: the MJ SACD produces overs even when 'directly' converted at +0dB. That suggests those tracks were mastered out of SACD spec.

Fortunately these particular overs don't sound as clicks. Or anything, on my system. Probably they are too isolated and brief.
 
Last edited:

ssully

2K Club - QQ Super Nova
Since 2002/2003
Joined
Jul 2, 2003
Messages
3,380
Location
in your face
Very interesting about the SACD peaks on some tracks.

I have a rip of the SACD and hear very audible problems on some of the kick and snare hits at the beginning of 'Billy Jean' using foobar with foo_dsd_asio and a Topping DSD DAC. It sounds like a popping sound and/or a bit of phase cancellation. It's not subtle at all and I figured maybe a better DAC might handle them better or possibly there's some settings I could change somewhere. This is the only track I've noticed problems on out of many dozens of SACD rips.


It may be that it was ripped too hot. This one has to be ripped with no level boost at all.
 

SeeMoreDigital

Well-known Member
Joined
Oct 23, 2018
Messages
181
Location
UK
Ripping the dsd bit-stream from an SACD does not alter any levels. It should be a 1:1 copy, minus any encryption.

I guess re-encoding a dsd bit-stream to an LPCM bit-steam could introduce errors (depending on its sample-rate and bit-depth), as could playing the LPCM bit-steam in any given software or hardware SoC LPCM decoder...
 

ssully

2K Club - QQ Super Nova
Since 2002/2003
Joined
Jul 2, 2003
Messages
3,380
Location
in your face
Ripping the dsd bit-stream from an SACD does not alter any levels. It should be a 1:1 copy, minus any encryption.


True. Ripping alone is 1:1.

I guess re-encoding a dsd bit-stream to an LPCM bit-steam could introduce errors (depending on its sample-rate and bit-depth), as could playing the LPCM bit-steam in any given software or hardware SoC LPCM decoder...


There is no question. Digital over "errors" can be introduced if level increase is purposely incorporated into a rip+PCM convert workflow by the user. Or if it is automatically employed by a player that outputs PCM from DSD. Sample rate and bit depth have nothing to do with it in the normal run of things. DSD was designed to be safely converted to 44/16 bit audio. If done correctly, DSD-->PCM should not introduce any overs, much less the ones seen on Thriller.

The errors 'baked in' to the Thriller SACD are not due to consumer choices or consumer hardware/software processing. They occurred during the SACD production process.
 

SeeMoreDigital

Well-known Member
Joined
Oct 23, 2018
Messages
181
Location
UK
DSD64 was designed to be safely converted to 44/16 bit audio. If done correctly, DSD-->PCM should not introduce any overs, much less the ones seen on Thriller.
Indeed... If done correctly. Which is why 44.1KHz steps are preferable, ie: 88.2KHz and 176.4KHz for DSD64 (2'822'400 snippets of data) disc to PCM re-encodes.
 

ssully

2K Club - QQ Super Nova
Since 2002/2003
Joined
Jul 2, 2003
Messages
3,380
Location
in your face
I'm not sure what your point is.

The absurdly high sample rates of DSD 128,256,521 are also integral multiples of CD rate (44.1). So they too inherit the designed-in 'intent' to easily render to 16/44 PCM. There is nothing special about them in that regard and no need to correct my original post to 'DSD64'.

I personally know of no incorrect DSD-->PCM results in consumer software or hardware, that weren't due to incorrect level increase settings. If you know of any I'm curious to hear more about them.

It is simple to guarantee that conversion will not *introduce* overs: convert @ +0 dB setting, i.e., no level increase.

Thriller's handful of overs persist at this setting too. The 'incorrectness' is not due to incorrect conversion.
 
Top