Skip to main content

Topic: Avoiding dynamic range compression...? (Read 7606 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
Avoiding dynamic range compression...?
Hi,
Well this is my first post here and I've gotta apologise if this topic has already come up elsewhere. [I didn't find anything on the interweb, which is why I'm asking  =]

Basically I want to avoid a problem which is increasingly well-documented on the internet and, when you examine it in detail, completely stupid. - As I understand it (correct me if I'm wrong on anything) a CD is technically capable of holding audio of a far better dynamic range and quality than all its predecessors - but virtually all record companies are forcing sound engineers to make the record "louder" by compressing it, causing clipping and the general replacement of music with static.
"The quiet bits sound loud and the loud bits just sound like noise"

Eg. Oasis. I'm a big fan and I love their music, but some of their albums have been so badly compressed it sounds like my speakers have blown and it gives you a headache
That's really, really not cool. CD theoretically could be perfect for my needs, but this seems to be ruining everything that should be fine



....So anyways. If I want to avoid this, what are the reasonably-high-quality alternatives to a CD? Vinyl is the decidely retro first choice of snobs audiophiles, but is it worth ripping vinyl albums to a lossless format on PC/iPod for any noticable improvment - and what about alleged distortion from dust etc, is that a big deal?

I guess the more modern alternative technologies are SACD and DVD-Audio.  But again, I'm afraid I'm an ill informed noob.....
- Are most vaguely popular albums available, what are the pros and cons of each format and how, if at all, can I avoid my other pet hate of DRM?
(oh and internet downloads are out, at least till we have Gbps connections and very high quality files..... by definition, if I could download my entire music library, it'd be in a format so compressed that a CD would be better anyways. =P)
Perhaps there's something else I've missed.... other than breaking into record studios worldwide and stealing their master copies


...If anybody can answer any of that, you've made my day.    Just need some advice on how best to go one better than CD, basically
- But surely other people are wondering the same thing....right.... ?

  • Axon
  • [*][*][*][*][*]
  • Members (Donating)
Avoiding dynamic range compression...?
Reply #1
Truthfully?

You can't avoid it, at least not generally. If you have specific knowledge that a particular release of an album was treated a different way than the others, you can use that to make an informed decision. Ie, if you know that the masters for an album are notably less compressed than the final CD, and that a vinyl/DVD-A/SACD release keeps more of that dynamic range than the CD, you're obviously safe.

But there are provable examples of high-res and vinyl releases having the same amount of dynamic range as the equivalent CD release. So saying that vinyl, or SACD, or DVD-A albums always have more dynamic range is absolutely false. Moreover, even if somebody claims that one format sounds better than another for a particular album, for all you know, it might be placebo.

You either need to find some numeric result to show that there is more dynamic range in one release vs another, or you need to hear from the horse's mouth (ie the mastering engineer) that less compression/limiting was applied. The latter is extremely hard to find, and the former requires somebody to have already bought the album in both formats, and to have compared the two in an audio editor.

  • knutinh
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #2
What do you think about this concept?
http://www.digido.com/misc-content/honor-roll.html

Quote
This is a collection of well-mastered pop CDs that have good micro (and macro) dynamic content. Microdynamics contain the transient movement or dynamic rhythm of a recording. Transients also affect perceived loudness. Macrodynamics contain the variations in average loudness over time. As we move into higher resolution recordings, I hope that DVDs, DVD-As, and SACDs will make it into this list. I've organized the recordings in order of increasing intrinsic loudness. Intrinsic loudness is the perceived loudness of a recording at a known position of the monitor control. So we can measure the difference in intrinsic loudness between two CDs by observing the difference between the two monitor positions required to make them sound equally loud.

Remember that we have no control over where the consumer sets his personal volume control, so the whole idea of "absolute loudness" is a fallacy. That is why I've introduced the term "intrinsic loudness" to help us compare recordings at a given known position of the calibrated monitor level control. Also note that high intrinsic loudness necessarily produces lowered microdynamics because of the compression/limiting required to prevent peak overload.

To me, a comprehensive database of measured dynamics alowing me to avoid the worst stuff would be idea.

-k

  • Axon
  • [*][*][*][*][*]
  • Members (Donating)
Avoiding dynamic range compression...?
Reply #3
That approach only applies for comparison of different albums on CD, not different masterings on different formats.

How accurate would that metric be for vinyl? Normalizing a digital transcription of a vinyl album against a CD counterpart is a nonstandardized process, and any absolute RMS metric is going to be corrupted by that.

  • MichaelW
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #4
(oh and internet downloads are out, at least till we have Gbps connections and very high quality files..... by definition, if I could download my entire music library, it'd be in a format so compressed that a CD would be better anyways. =P)


What everybody said, and if you search on "Loudness wars" (or just look for threads on that topic) you'll find much info.

Just one point: from the bit I've quoted, it seems possible that you haven't quite got the distinction between dynamic range compression (= loudness wars, when excessive, and is evil) and file size compression.

Search on "transparency" for the details, but the vast majority of people can enjoy the vast majority of music for the vast majority of the time with quite high ratios of (lossy) file size compression. The two things called "compression" are not intrinsically linked.

The whole world knows just what you're talking about, but the majors are in panic and quite irrational.

Michael

When in danger, fear and doubt,
Run in circles, scream and shout.

  • dcorban
  • [*]
Avoiding dynamic range compression...?
Reply #5
It's interesting that you mention Oasis. Out of many compressed albums I have, Oasis (What's the Story...) is by far the most compressed album I own. It is unlistenable, which is a shame.
Who controls the past controls the future: who controls the present controls the past

  • slks
  • [*][*][*][*]
Avoiding dynamic range compression...?
Reply #6
There's no surefire way to avoid it. Overcompression can be (and usually is) applied regardless of format.
  • Last Edit: 04 January, 2008, 11:35:45 PM by slks

Avoiding dynamic range compression...?
Reply #7
....Thanks everybody for the replies, first of all 

So am I to understand that there is usually no difference in 'quality' between CDs, vinyl or anything else because they are all mastered more-or-less the same?
Hmmmm. I can believe that the industry would do that, but if that's true why on earth do people always talk about vinyl sounding much more natural and what would be the point to SACD type technologies?
- About 6 months ago in the audio shop down the road from here, I heard Keane's Hopes and Fears first on CD and then on vinyl. Admittedly some of this difference was probably caused by the change in systems [the first was an Arcam seperate and the second, well I can't remember but suffice to say it was priced at about £10,000!], but even so the additional depth to the album on the record was staggering.....

Oh and @the (What's the Story) Morning Glory post: in my experience, that's one of their least compressed albums. Don't Believe the Truth is a lot worse and that's saying something
  • Last Edit: 05 January, 2008, 06:30:46 AM by Joey Jo Jo Jr. Shabadoo

  • knutinh
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #8
That approach only applies for comparison of different albums on CD, not different masterings on different formats.

How accurate would that metric be for vinyl? Normalizing a digital transcription of a vinyl album against a CD counterpart is a nonstandardized process, and any absolute RMS metric is going to be corrupted by that.

Comparing peak amplitude to "average" amplitude should be possible both algorithmically and perseptually, no matter if the system peak amplitude is changing.

If a CD and a vinyl record is played back and a human operator is asked to control the level until they are "equally loud", then it is simple to measure max and rms levels.

-k

  • MichaelW
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #9
Ignorant question. I remember way back when, a friend who was serious about music but not an audiophool had a thing called a dynamic range expander, to cope with the compression needed for vinyl.

I realise that nothing could be done about information lost in clipping, but would it be possible to do something digital to make quiet passages quieter?

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
Avoiding dynamic range compression...?
Reply #10
You have to match the compression curve in order for expansion to give a decent result but it's even more complicated than that due to thresholds, attack and release times and other parameters.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

  • MichaelW
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #11
You have to match the compression curve in order for expansion to give a decent result but it's even more complicated than that due to thresholds, attack and release times and other parameters.


Thank you.

Given that modern lossy compression is so good, and the loudness wars are a Great Scourge, would expansion be the next great challenge for all the dedicated people who've worked on Lame, Ogg Vorbis and all the others, or is the information hopelessly lost?

The sort of quality that would be an improvement for me would be like this: as an old fart, I'm buying Greatest Hits CDs to recover the sound track of my youth. These are, more often than not, so compressed as to be unenjoyable, but I remember having listened to them with delight on cassette or even a crappy gramophone (not even stereo, necessarily). Perfection not necessary, just loud bits and quiet bits.

Michael

who *would* try it himself, if he had the knowledge, and could be induced to put up a small amount of cash for a bona fide effort

  • Bourne
  • [*][*][*][*][*]
  • Banned
Avoiding dynamic range compression...?
Reply #12
anyone has got OASIS/WTSMG in vinyl format so that we can make a comparison between the LP and CD loudness????

  • pdq
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #13
You have to match the compression curve in order for expansion to give a decent result but it's even more complicated than that due to thresholds, attack and release times and other parameters.

And that's just to counter automated compression of some kind. There can also be manual manipulation of the levels and inappropriate mixing etc. There can be no way to automate the correction of these kinds of compression. On the other hand, even inaccurate expansion could make really awful tracks a little more listenable.

  • krabapple
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #14
It's interesting that you mention Oasis. Out of many compressed albums I have, Oasis (What's the Story...) is by far the most compressed album I own. It is unlistenable, which is a shame.



That album was released on SACD too. It would be interesting to see if the mastering differs from the CD's.

Btw, for the OP, digitizing vinyl doesn't afford any automatic sonic improvement, but you can go in and
fix pops, clicks, weird EQ, on the digital file.  And digitized audio is invulnerable to the sorts of degradation an LP can suffer.





....Thanks everybody for the replies, first of all 

So am I to understand that there is usually no difference in 'quality' between CDs, vinyl or anything else because they are all mastered more-or-less the same?


No, that would be a misreading of what's been written.

Quote
Hmmmm. I can believe that the industry would do that, but if that's true why on earth do people always talk about vinyl sounding much more natural and what would be the point to SACD type technologies?


Reread again -- an LP mastering MAY  (or may not) have more dynamic range than the CD version.  You have to do the actual measurements to be sure.  You can get higher levels of compression on a CD, but you can still get high compression on an LP.  With SACD, though there are actual 'spec' limits on SACD mastering that discourage mastering moves that would result in clipping, but compression can be applied in the PCM (CD) realm before transcoding to DSD (the SACD format).  So again, to know the relative levels, of SACD to CD, you'd have to do the measurements.  You can't assume.

As for the 'point' of SACD, some would argue that it was a means to enforce copy protection...SACD still has the best DRM of any of the hard media formats.  It's never been cracked, AFAIK.

Wikipedia has a very good (not perfect, but very good) article on SACD.

http://en.wikipedia.org/wiki/Super_Audio_CD


Quote
- About 6 months ago in the audio shop down the road from here, I heard Keane's Hopes and Fears first on CD and then on vinyl. Admittedly some of this difference was probably caused by the change in systems [the first was an Arcam seperate and the second, well I can't remember but suffice to say it was priced at about £10,000!], but even so the additional depth to the album on the record was staggering.....

Oh and @the (What's the Story) Morning Glory post: in my experience, that's one of their least compressed albums. Don't Believe the Truth is a lot worse and that's saying something



The earlier Oasis album was one of the 'watershed' CDs in the loudness wars.  Compared to what came before, it was radical.  Compared to what came after, it was conservative.
  • Last Edit: 05 January, 2008, 11:08:18 PM by krabapple

  • Bourne
  • [*][*][*][*][*]
  • Banned
Avoiding dynamic range compression...?
Reply #15
WTSMG is really bad... It's unlistenable... soon I find out that there's not a lot of Oasis music that can be heard decently. Probably the live album Familiar to Millions will sound a bit softer, but the rest is a absurdly annoying. I remember a chick that would tell me everytime she heard an Oasis song she would get a headache, at that time I still didn't know what caused that....

  • knutinh
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #16
Given that modern lossy compression is so good, and the loudness wars are a Great Scourge, would expansion be the next great challenge for all the dedicated people who've worked on Lame, Ogg Vorbis and all the others, or is the information hopelessly lost?

Persuading record companies into separating the information into "raw mix" and "master enhancements" would be more feasible and probably still impossible. I am talking about a "linear" track (or the degree of linearity that a recording engineer would like in an ideal world with noise-less, dynamic playback systems and no executives hanging over their shoulder), accompanied by a difference-track or compressor parameters allowing flat distorted loudness for those that prefers it (such as radio stations).

I think that the movie industry has taken sound quality lot more serious than the recording industry. That is a paradox. They have norms for absolute referenced playback-level, dynamic headroom, user-selectable dynamic compression, playback acoustics and speakers etc.

It is a shame that the recording industry operate on the single parameter of maximum perceived loudness within a peak-amplitude limited channel.

It is also a shame if the recording format has to be "crippled" enocoding-wise to refuse recording engineers making bad-sounding albums...


Some dynamic processors are multi-band, combining "EQ" with "compression" in a complex fashion. Other are modelled by simple tube units with very few parameters. As an academic exercise, it would be interesting to see what can be done to a compressed track if the compressor is known, and perhaps a few "standard" settings are suggested into some recurssive algorithm.

-k
  • Last Edit: 06 January, 2008, 12:06:24 PM by knutinh

  • 2Bdecided
  • [*][*][*][*][*]
  • Developer
Avoiding dynamic range compression...?
Reply #17
What do you think about this concept?
http://www.digido.com/misc-content/honor-roll.html

Excellent.

"K-14" = -6dB on Bob's scale = 0dB on the current ReplayGain scale.
"K-20" = 0dB on Bob's scale = +6dB in ReplayGain.


Apart from the obvious defects, and various limitations that are completely inaudible when replaying "loudness war" content, there is nothing to choose between CD, LP, DVD-A and SACD in terms of dynamic range.

What is somewhat common is that an intentionally less "smashed" version is released on LP or SACD, making it sound superior to the CD. This is because it's a different signal being put onto the disc, not due to any property of the format.

Be careful though: often all formats are mastered with exactly the same signal. It's important to get some verification that the vinyl or SACD release contains a superior version before wasting your money.

Also, I don't find vinyl or SACD convenient - they should be copied onto CD, and/or converted to mp3!

Cheers,
David.

  • knutinh
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #18
I think it is important to re-iterate what you are saying, because so many consumers and magazines are confused.

1. There is no evidence that CD,  vinyl, DVD-A, SACD etc have inherent dynamical limitations that matters on a perceptual scale. On a technical scale, in order of increasing available format dynamic range would be [vinyl, CD, SACD, DVD-A].

2. There are many anecdotal evidence that the content is different in many but not all cases. This means that a vinyl recording may easily have better dynamics (or worse) than a CD, without the format having anything to do with it.

3. Therefore, any report or opinion about CD vs SACD that is only based on A/B-ing between e.g. the different layers of an SACD is of little value for explaining format-only limitations.

4. A user may, however, conclude that the empirical quality of available SACDs (i.e. format + mastering-engineer) are sufficiently better than CDs to warrant a system-change.

-k

Excellent.

"K-14" = -6dB on Bob's scale = 0dB on the current ReplayGain scale.
"K-20" = 0dB on Bob's scale = +6dB in ReplayGain.

But it is interesting to publish those numbers, something that strangely seldom happens with RG.

I admit that there is no -1:1 correlation between the joy that an album can bring and its perceived loudness (not even between percieved audio quality and perceived loudness), but being able to rule out some bad ones before purchasing (and leveling out using RG), would be a nice possibility. It would also send a clearer signal to the bean-counters at record companies...


It is also interesting if the "all-perseptual" and the "techie/perseptual" approach converge. All to often, the two camps are too busy bad-mouting each other.

-k
  • Last Edit: 07 January, 2008, 08:26:26 AM by knutinh

  • Axon
  • [*][*][*][*][*]
  • Members (Donating)
Avoiding dynamic range compression...?
Reply #19
The placebo effect still applies for comparisons of different masterings, so I would suggest that #4 is unsafe to rely on for evaluating masters.

Also, isn't a loudness estimator a fundamentally unsound metric for dynamic range? I mean, dynamic range isn't even what ReplayGain trying to measure. Clearly, loudness defines a maximum dynamic range, but pop music clearly doesn't hit that maximum consistently. Very few albums have reached Merzbow levels.

[!--sizeo:1--][span style=\"font-size:8pt;line-height:100%\"][!--/sizeo--](And yes, I have a better idea, but I'm waiting on procuring a download location for it before discussing it.)[/size]


Quote
or you need to hear from the horse's mouth (ie the mastering engineer) that less compression/limiting was applied. The latter is extremely hard to find
Just to toot my own horn a little bit longer by quoting myself, an unusually good example of this is Steve Hoffman's commentary on his mastering projects. This sort of information is absolute gold when evaluating different releases, although you still may not have much of an idea how audible the differences are.
  • Last Edit: 07 January, 2008, 01:20:13 PM by Axon

  • DVDdoug
  • [*][*][*][*][*]
Avoiding dynamic range compression...?
Reply #20
Quote
Ignorant question. I remember way back when, a friend who was serious about music but not an audiophool had a thing called a dynamic range expander, to cope with the compression needed for vinyl.

I realise that nothing could be done about information lost in clipping, but would it be possible to do something digital to make quiet passages quieter?
  Even without clipping, information is lost...  There is no way to know if a 0db peak was +3dB or +9dB before compressing/limiting.  So, even if you did know the compression algorithm, it's not linear, and it's not an accurately-reversible process.

That said, YES!  Just about every audio editor has an expander.  You can generally set the expander for "downward expansion" of the quiet parts, and/or for "upward expansion" to boost the peaks.  Although you can't truly restore a compressed recording, you may be able to improve it somewhat.  Like everything else in audio, you can improve a slightly compressed recording, but if it's severely compressed you might end-up making things worse when you try to improve it.

And, although you cannot accurately restore a clipped peak, there is software that will make estimations & assumptions to recreate a peak.  There is even a FREE Declipper.
  • Last Edit: 07 January, 2008, 02:41:21 PM by DVDdoug

Re: Avoiding dynamic range compression...?
Reply #21
Hi folks, apologies for the ressurecting this ancient thread. I have very similar question regarding the current trend to compress the sound on a CD.

Basically, what I d'like to know is why there's a need to compress the sound before they transfer it to the CD?  I have read multiple CD vs SACD discussions and came to the conclusion that the good old CD is more than capable to deliver great sound.

Is this all about loudness was? Sound is being compressed to achieve louder record on a CD? Compression is being used to mask any artifacts or noises? What is the reason?

Thanks in advance!

  • DVDdoug
  • [*][*][*][*][*]
Re: Avoiding dynamic range compression...?
Reply #22
Quote
Is this all about loudness was? Sound is being compressed to achieve louder record on a CD? Compression is being used to mask any artifacts or noises? What is the reason?
Yes...  loudness (and consumer demand).   Due to the integer nature of CDs and digital-to-analog converters there is a peak limit.     Once the peaks are hitting the limit you can bring-up the loudness by compressing & bringing-up everything else.    But of course, you can always boost the analog gain if your system is capable of it...   I think the digital "loudness" is more of a first-impression issue.

If highly-dynamic recordings were to win a few Grammies and sell millions of copies there would be more highly-dynamic releases.   And,  I suspect that musicians are trying to play less-dynamically to sound like the music that's selling.   Constant-volume seems to be what the public wants.    :(

Since compression is used to bring-up the quieter parts, it tends to bring-up any background noise...  It generally makes noise worse. 

But, it does mask noise in in the playback system or in the medium...   If you've listened to vinyl, I'm sure you've noticed that the snap, crackle, and pop are more noticeable during quiet passages.   Or if you've listened to cassettes, you've probably noticed that tape hiss seems worse during quiet parts.    The old analog noise reduction systems (Dolby and DBX) worked by compression during playback (for a good signal-to-noise ratio) and then expansion during playback.

Re: Avoiding dynamic range compression...?
Reply #23
DVDdoug, thanks for answering.