Why are you asking about Sony when BlueNote is the responsible label?

Kal

Wish I had seen this sooner. Just got into SACD a few months ago. Bought this album and in SACD 2 channel was surprised that it didn't sound any better than my cd (played back on a different machine).
Now I know why. Will try comparing the Mch SACD layer down mixed to 2ch vs the 2 channel cd. Wonder if the mixed down hi-rez will sound better than mid-rez cd?
thanks
barondla

Quote:

Quote:
Bought this album and in SACD 2 channel was surprised that it didn't sound any better than my cd (played back on a different machine).
Now I know why.

This post probably confirms Blue Note has never replaced the SACD 2-track on this album with a high rez version. But I was hoping to get a more definite answer from JM or JA.

As far as I am aware, the SCD wasn't ever remastered.

Quote:
Perhaps more criticism from the audio press directed at artists and record producers would serve to curtail this deceptive practice.

All we can do is comment publicly when we discover examples.

Quote:
IMO the audio press should also more directly chastise artists and record producers when they dump badly recorded and/or engineered music albums on the market.

There are months when I despair of finding a good-sounding new release to be featured in our "Recording of the Month' section. Overcompression and even plain old distortion are ubiquitous, sadly.

John Atkinson
Editor, Stereophile

Very sad!
Do you often consider labels like Yarlung, Reference Recordings, Mapleshade, etc.? Please don't think I'm being like others who seem to be telling you how to do your job; I'm not that smart! lol Thanks for all your hard work John!!

Quote:
I would like to see the audio press become much more aggressive in calling out the badly recorded RBCD's being produced. Perhaps magazines such as Stereophile could write more editorials chastising record producers (by name/album). At the least I feel there should be significantly lower sound ratings for badly recorded albums than I see currently.

John Atkinson has discussed the damage done to music when dynamic contrast is intentionally removed to make recordings louder. For instance: Dynamics & Dynamic Range, Overcooked Floyd, and CD Quality: Where Did the Music Go?. And both JA and music editor Robert Baird return to the subject in our upcoming November issue.

Would it be possible to show some sort of measurement for the reviewed music?
After all measurements are presented for nearly every other aspect of reviews on Stereophile.
It would help to put the mastered tracks on the spot when their quality can be compared to other albums, especially if measurements spread to other music-audio magazines (long term ideal dream I guess).
So is it practical to show say average A-weighted RMS and raw RMS for the whole album, or say for best/worst tracks?
To begin with for most this may not have meaning, but once a listener can compare a few measured music CDs (as an example) a point of reference is created for those values to have relevance.
Example of a mastering engineer showing some albums with the said values.
http://mastering-media.blogspot.com/2008/10/diy-mastering-part-5-how-loud-is-too.html

Cheers
Orb

Quote:
Would it be possible to show some sort of measurement for the reviewed music?

(blink) What kind of measurement would you mean, now?

I can think of a couple of interesting measures for standard PCM (no, THD, SNR, etc need not apply since there is no "original" to compare to), but that's not what we're talking about here. Perhaps some of them could be approximated in SACD.

Quote:

Quote:
Would it be possible to show some sort of measurement for the reviewed music?

(blink) What kind of measurement would you mean, now?

I can think of a couple of interesting measures for standard PCM (no, THD, SNR, etc need not apply since there is no "original" to compare to), but that's not what we're talking about here. Perhaps some of them could be approximated in SACD.

No need to blink
Specifically what I mentioned that ties in with the multiple posts above mine on compressed dynamics.
I also included a URL from a mastering engineer showing the effects and providing a basic reference point for use of A-weighted RMS and raw RMS averaged measurements.
Also he answers a few points at the bottom of the blog.

His perspective and points seem to fit in with other mastering engineers.

I guess I am coming from the perspective of focusing where the greatest problem currently lies with audio quality for CD/SACD (but also applies to LP/etc, and this ties back to the mixing but more importantly IMO the mastering.

Cheers
Orb

Quote:

Specifically what I mentioned that ties in with the multiple posts above mine on compressed dynamics.
I also included a URL from a mastering engineer showing the effects and providing a basic reference point for use of A-weighted RMS and raw RMS averaged measurements.
Also he answers a few points at the bottom of the blog.

So, perhaps, one could even improve that by plotting a loudness measure (as opposed to some kind of intensity measure) as a function of time, or, if one could make a color graphic, do a partial loudness spectrogram. This would show dynamic range, frequency balance, etc, in terms that would relate better to human perception.

Perhaps also a histogram of loudness would be revealing.

For PCM signals, a histogram of the actual PCM levels (-32768 to +32767 is very revealing. I've found more than one CD in which the two most common levels were + and - max, and both were more likely than the zero level. I think the term for this is "clipped".

What's also insidious is that I've found CD's that show the same pattern (figure someting looking somewhere between log-normal and gaussian with a peak at -max and +max, but where "max" isn't PCM maximum, but is rather something between 16k and 32k. This does not show up on most clipping measurements, but is still clipped to (bleeeeeep). I've also found missing codes (one CD has every 7th negative level missing, starting with -4. you figure), others show clear gain changing after dithering (instead of before). All sorts of things show up in these plots.

And I am a charter member of the TURN IT DOWN club, so showing loudness plots and detecting potential bad processing is appealing.

I would think it has to be simple enough to fit in with the music reviews that have limited page content, hence why 2 lines with dBFS takes up a small amount of room and easy to reference against other albums.
Also, seems to me mastering engineers use this technique and as the URL I provided shows, the level distortion and clipping can be pretty easily judged with those figures.

Also there is the need to "bridge the chasm"; meaning the information has to reach not only early adopters but also those who are not technical savvy or do not want excessive data so its impact has greatest results.
Also using say a standard that already exists may help in convincing other magazines to adopt the measurements, but someone needs to take the 1st step and show it working in principle.

And yes, it is clipping as it seems you were not too sure, the link to the mastering engineer blog is pretty informative and worth a look.
http://mastering-media.blogspot.com/2008/10/diy-mastering-part-5-how-loud-is-too.html

Cheers
Orb

Quote:

And yes, it is clipping as it seems you were not too sure,

For a guy from the UK, you seem to miss understated humor. No kidding, it's clipping, duh. Egregious, unbelievable, astonishing levels of horrid clipping, to boot.

Yes, I'm aware of a lot of things going on in mastering. The reason I was proposing a graph is because it carries a lot more information than a single weighted RMS level. I suppose that a weighted RMS level is better than nothing. Thing is, I can get a weighted RMS level that looks reasonable by having one song way too quiet and the next way too loud. A graph would expose this. Using loudness instead of a weighted RMS level would also actually extend the number to something one actually experiences. Something at -10dB FS (peak) at 50 Hz is just not the same as sine tones at 800, 1200, 1800, and 2400 Hz, each of equal power, totalling the same power as the 50Hz sine wave, to put it mildly.

Didn't the Boston Audiophile Society prove that you can't hear the difference between CD and SACD, anyway?

ahem...DSD has its place in one (1) application: Mastering from analogue. When the recording chain is completely analogue, you can feed the audio from the analogue mastering into a DSD A/D converter and cut that signal straight onto a disc without any further processing. It is in this application that DSD can be viewed as pretty transparent. When you convert the signal twice, however (such as when using a DSD recorder as the tracking medium), the second conversion is no longer transparent, due to the HF noise present in the source signal hitting a second analogue deltasigma modulator.

Quote:
I've also found missing codes (one CD has every 7th negative level missing, starting with -4. you figure)...

Well, if it's an old CD, missing codes don't sound all that unlikely. The old successive-approximation A/D converters, when implemented with a non-monotonic D/A in the feedback loop, will have missing codes. Back in the old days, it was actually pretty hard to find a D/A that was monotonic to 16 bits.

If it's a new CD, I dunno how to explain it.

Quote:

Quote:

And yes, it is clipping as it seems you were not too sure,

For a guy from the UK, you seem to miss understated humor. No kidding, it's clipping, duh. Egregious, unbelievable, astonishing levels of horrid clipping, to boot.

Yes, I'm aware of a lot of things going on in mastering. The reason I was proposing a graph is because it carries a lot more information than a single weighted RMS level. I suppose that a weighted RMS level is better than nothing. Thing is, I can get a weighted RMS level that looks reasonable by having one song way too quiet and the next way too loud. A graph would expose this. Using loudness instead of a weighted RMS level would also actually extend the number to something one actually experiences. Something at -10dB FS (peak) at 50 Hz is just not the same as sine tones at 800, 1200, 1800, and 2400 Hz, each of equal power, totalling the same power as the 50Hz sine wave, to put it mildly.

Looks like I was wrong giving you the benefit of doubt of not always being sarcastic, of course if I had retorted I am sure you would had said I was biased.
Due diligance (I could had assumed you were intentionally being sarcastic but then this ignores the potential of you actually being serious) and open mind James, try it on these forums once in awhile mate

If one song is way too low and the next is way too high, you show the lowest track and also loudest.
However realistically have you come across that many albums where tracks from an artist are that far spread apart?
And using a chart does not help bridge the chasm with those who just do not care for the same technical detail as yourself.
I understand your thoughts about the variability of dBFS of tracks on an album, but your chart would have to do the same.
You may not feel its ideal but mastering engineers seem happy to use it, even to the extent to discuss it on blogs and show actual albums with long term A-weighted and raw dBFS levels.

Using your example, does it come out different to the link and the discussion point raised by an actual mastering engineer who makes a living doing major records?

Cheers
Orb

Cheers
Orb

Quote:
Didn't the Boston Audiophile Society prove that you can't hear the difference between CD and SACD, anyway?

It has shown that for a not-too-bad test setup with pretty good equipment, you can't hear the difference unless you have a gain structure that exposes the noise present during silence.

That's what it shows.

And the noise present in silence was in fact noted by listeners to a very confident degree, so the test couldn't have been THAT horrid.

But it also did show that you could make a distinction, so your summary is a bit overstated.

Quote:
If one song is way too low and the next is way too high, you show the lowest track and also loudest.
However realistically have you come across that many albums where tracks from an artist are that far spread apart?
And using a chart does not help bridge the chasm with those who just do not care for the same technical detail as yourself.
I understand your thoughts about the variability of dBFS of tracks on an album, but your chart would have to do the same.
You may not feel its ideal but mastering engineers seem happy to use it, even to the extent to discuss it on blogs and show actual albums with long term A-weighted and raw dBFS levels.

Using your example, does it come out different to the link and the discussion point raised by an actual mastering engineer who makes a living doing major records?

Cheers
Orb

Cheers
Orb

What do I expect? Somebody, somewhere, will try very, very hard to "work" any kind of measure that is proposed. The quiet/hot scenario is likely to be seen soon in the USA, I fear, in some broadcast situations, thanks to some standards that appear to be pending, and again thanks to using a mean or median, rather than a distribution, in order to make a measurement.

I think it would, under most (but not all) situations be odd to make a pop recording that way, but consider Beethoven's 6th. It would actually make sense there.

So a time vs. loudness plot would show you what you get. It doesn't have to be enormously detailed, and could be plotted on a 30 second per tick axis and still carry meaningful information. This could reduce to a standard height plot in print without losing the intended information, and is something that I think most any reader could understand.

Quote:

Quote:
I've also found missing codes (one CD has every 7th negative level missing, starting with -4. you figure)...

Well, if it's an old CD, missing codes don't sound all that unlikely. The old successive-approximation A/D converters, when implemented with a non-monotonic D/A in the feedback loop, will have missing codes. Back in the old days, it was actually pretty hard to find a D/A that was monotonic to 16 bits.

If it's a new CD, I dunno how to explain it.

But even SAR missing codes wouldn't have a pattern like every 7th missing code. The only way I know how to get that is to use a multiply by some number close to 1, and then fail to dither, usign 16 bit arithmetic.

One would HOPE nobody would do that. Yeah. I know. HOPE springs aeternal and all that, too.

If you have access to matlab or octave, suck in a PCM track and analyze it. You may wish to hold your nose if it's a modern ultracompressed track, however.

Quote:

Quote:
If one song is way too low and the next is way too high, you show the lowest track and also loudest.
However realistically have you come across that many albums where tracks from an artist are that far spread apart?
And using a chart does not help bridge the chasm with those who just do not care for the same technical detail as yourself.
I understand your thoughts about the variability of dBFS of tracks on an album, but your chart would have to do the same.
You may not feel its ideal but mastering engineers seem happy to use it, even to the extent to discuss it on blogs and show actual albums with long term A-weighted and raw dBFS levels.

Using your example, does it come out different to the link and the discussion point raised by an actual mastering engineer who makes a living doing major records?

Cheers
Orb

Cheers
Orb

What do I expect? Somebody, somewhere, will try very, very hard to "work" any kind of measure that is proposed. The quiet/hot scenario is likely to be seen soon in the USA, I fear, in some broadcast situations, thanks to some standards that appear to be pending, and again thanks to using a mean or median, rather than a distribution, in order to make a measurement.

I think it would, under most (but not all) situations be odd to make a pop recording that way, but consider Beethoven's 6th. It would actually make sense there.

So a time vs. loudness plot would show you what you get. It doesn't have to be enormously detailed, and could be plotted on a 30 second per tick axis and still carry meaningful information. This could reduce to a standard height plot in print without losing the intended information, and is something that I think most any reader could understand.

Then you really should take this up with mastering engineers, have you discussed this with Bob Katz at AES?
Interested to know what was said or if you have had the opportunity to discuss this subject with him.

Just so it is clear please can you expand on how the 30 second tick chart would look like for say the following:

Quote:

-8.6 (-6.2) Oasis - "Some Might Say": Severe clipping distortion
-8.9 (-4.9) Metallica - "TDTNC" (CD): Massive distortion & clipping
-10.4 (-7.7) Feeder - "Pushing The Senses": Heavy clipping distortion
-12.7 (-7.7) Metallica - "BB&S" (Mystery Mix): Slight source clipping
-14.0 (-10) Katatonia - "Consternation": Awesome (clean) sound, massive choruses
-15.3 (-13.1) Sugar - "Fortune Teller": From 1993
-21.8 (-16.9) Metallica - "TDTNC" (GH3) Needs to be louder !

Not sure how accurate the 30 second tick would be if the dynamics swing a lot over 10 seconds with another 20 following seconds being pretty flat.
The long weighted RMS calculations takes into account the whole of a track with accuracy down under a second.

I appreciat I may be missing something with the chart aspect, in my mind is accuracy use of the chart and the length of it when considering 4minute tracks and a complete album.

Edit:
Is this also compounded by the potential of many channels used in the recording studio with different levels assigned to each channel such as voice and a specific electric guitar(talking before downmixed to 2-channel for CD if this makes sense)?

Edit:
Ah forgot to say, there are pros and cons of either A-weighted or C-weighted for mastering, this is why I said to show both A-weighted and raw RMS figures.

James, this was all explained on that mastering blog URL and how the raw/A-weighted fits together in terms of listening for some tracks.

Cheers
Orb

Quote:

Quote:
Didn't the Boston Audiophile Society prove that you can't hear the difference between CD and SACD, anyway?

It has shown that for a not-too-bad test setup with pretty good equipment, you can't hear the difference unless you have a gain structure that exposes the noise present during silence.

That's what it shows.

And the noise present in silence was in fact noted by listeners to a very confident degree, so the test couldn't have been THAT horrid.

But it also did show that you could make a distinction, so your summary is a bit overstated.

It was sarcasm. My sarcasm was overstated.

Quote:

Quote:

Quote:
Didn't the Boston Audiophile Society prove that you can't hear the difference between CD and SACD, anyway?

It has shown that for a not-too-bad test setup with pretty good equipment, you can't hear the difference unless you have a gain structure that exposes the noise present during silence.

That's what it shows.

And the noise present in silence was in fact noted by listeners to a very confident degree, so the test couldn't have been THAT horrid.

But it also did show that you could make a distinction, so your summary is a bit overstated.

It was sarcasm. My sarcasm was overstated.

Ok, if you say so.

Quote:
James, this was all explained on that mastering blog URL

And you're still addressing me rudely, and not dealing with the difference between loudness and intensity.

As to "what would it look like" if you have all the tools you appear to have, why not make your own chart? I'm not going to go out and buy a half-dozen clipped-to-death albums that I will never ever listen to again just to measure them. I've measured enough horrible stuff already.

I guess I might reconsider if you would stop being rude when you address me, at least as far as showing a plot of loudness vs. time for something relevant, but something I already own.

ETA: http://www.fudco.com/chip/deconstr.html reminds me of what I seem to be arguing with. Jacques Derrida never did answer why the rock still falls on his foot.

the TT Dynamic Range meter will horrify rather quickly..

analyze a .wav from the 70s.

ok..now analyze a .wav from the 90s and 00s.. horrid!

http://www.audiogeekzine.com/?p=1064

http://www.pleasurizemusic.com/

Quote:
the TT Dynamic Range meter will horrify rather quickly..

analyze a .wav from the 70s.

ok..now analyze a .wav from the 90s and 00s.. horrid!

http://www.audiogeekzine.com/?p=1064

http://www.pleasurizemusic.com/

I think what can complicate this is if you do not analyse the whole track due to instruments on different channels at time of recording having different levels (could be very wrong here maybe JA can explain more on his experience).
What I mean is that listening/analysing a passage may show it to have flat dynamics while a later segment on the same track has large swings; whether this is down to channel settings at time of mixing or something else (such as at time of mastering or potentially combination of both) maybe someone can explain.

Also from what I can tell some non-pro software do not fully sample, whether it is only a portion of the track or the interval of sampling - just going by what I have seen others post.

Here is more measurement software recommend by Ian Shepherd, I appreciate you may enjoy your own so just throwing it out for those who may want to look at others:
http://mastering-media.blogspot.com/2008/05/mastering-software-1-metering.html
In the past I have used Wavelab (which also has the Bob Katz K-system metering as well).

Cheers
Orb

Quote:

Quote:
James, this was all explained on that mastering blog URL

And you're still addressing me rudely, and not dealing with the difference between loudness and intensity.

As to "what would it look like" if you have all the tools you appear to have, why not make your own chart? I'm not going to go out and buy a half-dozen clipped-to-death albums that I will never ever listen to again just to measure them. I've measured enough horrible stuff already.

I guess I might reconsider if you would stop being rude when you address me, at least as far as showing a plot of loudness vs. time for something relevant, but something I already own.

ETA: http://www.fudco.com/chip/deconstr.html reminds me of what I seem to be arguing with. Jacques Derrida never did answer why the rock still falls on his foot.

Sigh,
last post responding.
I have given you so many effing chances to be a grown up and yet your still being a complete arse.
Maybe one day you will lose your attitude and your one-sided bias.
As a suggestion actually try reading Geoffrey Moore "Bridging the Chasm"; for myself and others who are technology champions in leading edge companies this is usually highlighted as essential reading.

Also as a suggestion go back and read the debiasing article, I think it will help you....
Or probably not.

Orb

Quote:
the TT Dynamic Range meter will horrify rather quickly..

analyze a .wav from the 70s.

ok..now analyze a .wav from the 90s and 00s.. horrid!

http://www.audiogeekzine.com/?p=1064

http://www.pleasurizemusic.com/

The results certainly will horrify you, and on this we agree.

Quote:

I have given you so many effing chances to be a grown up and yet your still being a complete arse.

I see, so calling somebody a name they object to being called is "giving you so many effing chances to be a grown up"?

Sorry, that claim of yours makes you an overt fraud.

ONLY, and I repeat ONLY someone who wishes to be intentionally, deliberately, and obscenely rude will call someone by a name they do not like.

Ergo, you are a rude, intentionally offensive individual whose entire goal appears to be disparagement.

When you learn the veriest simple basics of human interaction, try again.

I use Sequoia and its little brother, Samplitude (v10, waiting to upgrade to v11).. they both come with the K System protocols in place(as a template option), though I dont really use it at all..

the TT meter analyzes either song-size or album-size chunks.... anything less is not meaningful..