Skip to main content

Topic: Bob Katz - Loudness: War & Peace (Read 9154 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • 2Bdecided
  • [*][*][*][*][*]
  • Developer
Bob Katz - Loudness: War & Peace
http://www.youtube.com/watch?v=u9Fb3rWNWDA

Or in Quicktime on Bob's site...
http://www.digido.com/loudness-war-explained.html
...with a downloadable version, and links to several papers on the topic.

Cheers,
David.
  • Last Edit: 18 November, 2011, 01:14:14 PM by 2Bdecided

  • krabapple
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #1
Am I misunderstanding, or does Katz get ADC and DAC confused in his description of analog clipping at around 6:40?
  • Last Edit: 18 November, 2011, 03:40:45 PM by krabapple

  • 2Bdecided
  • [*][*][*][*][*]
  • Developer
Bob Katz - Loudness: War & Peace
Reply #2
Am I misunderstanding, or does Katz get ADC and DAC confused in his description of analog clipping at around 6:40?
Yes, I think so. He explains the same thing, correctly, later.


This is an excellent video - the history part is very interesting.

When he seems to be saying is that by 2020 he thinks everything will be using something like ReplayGain by default (or in the media), and the only problem left will be that things that should be quiet, won't be.


He believes that 90% of the loudness war is due to producers wanting things louder, and 10% is due to wanting "that sound". I'm not convinced on this point, but I guess there are degrees of it. I don't think any pop music sounds "right" to modern ears without compression. It was compressed in the 1960s of course, but to a different degree. I think nearly half of the reason for what we hear today is that people want "that sound" - and they'd go for that sound even if every player in the land reduced the gain by 14dB to remove all of the loudness advantage.

Roll on the day of loudness-normalised media. I guess Apple will do it first

Cheers,
David.

  • krabapple
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #3
I agree.  I would love for 'more louder'  to be implemented on the playback side rather than coded into the music itself.  But peak normalization doesn't get you the same 'sound' or 'slam' as compression/limiting/clipping, so i wonder if Katz is being over-optimistic. Alternately, playback devices could include an option to apply dynamic range compression -- this is a common option on AVRs now, intended mainly for movie soundtracks.
  • Last Edit: 18 November, 2011, 05:15:12 PM by krabapple

  • Bob Katz
  • [*]
Bob Katz - Loudness: War & Peace
Reply #4
Am I misunderstanding, or does Katz get ADC and DAC confused in his description of analog clipping at around 6:40?



Dear Krabapple: You are absolutely correct. What I said and what I meant were two different things. I even knew that as I was editing the speech but it became too difficult to edit around my mis-speak and I simply decided to leave it in as the intent was clear. Go to the head of the class  :-).


Best wishes,



Bob

  • Bob Katz
  • [*]
Bob Katz - Loudness: War & Peace
Reply #5
I agree.  I would love for 'more louder'  to be implemented on the playback side rather than coded into the music itself.  But peak normalization doesn't get you the same 'sound' or 'slam' as compression/limiting/clipping, so i wonder if Katz is being over-optimistic. Alternately, playback devices could include an option to apply dynamic range compression -- this is a common option on AVRs now, intended mainly for movie soundtracks.


Dear Krabapple: Of course peak normalization doesn't get you the "slam" as heavy compression. But the key to understanding the loudness war is that if you apply peak normalization (which is a form of "gain makeup") AFTER applying your heavy compression and/or limiting, the loudness increase (and unfair advantage) is tremendous. Unfortunately, "more louder" will not be implemented on the playback side with the loudness-normalization schemes that are moving into place. And in fact it will be somewhat a reverse, as material that wants to sound louder (e.g. Beethoven's Ninth and Red Hot Chili Peppers) will have the same loudness as Joan Baez. But the situation will still be far more fair than the current scheme of "compress the Bejesus out of it and then peak normalize". And it will allow people who want to squash to coexist with people who want a more open dynamic. I'd like to think (hope) that will discourage program producers for compressing just for the sake of getting a loudness advantage. And I also predict that those of you who are big fans of slamming will discover (much to your psychoacoustic amazement) the great deal of influence that the loudness of the material has on your own preference! I've done the listening tests, loudness-matched all the zippy/slammed material you can put at me, and when it is loudness matched to dynamic versions of the same material, guess which ones usually win? 

Well, whether you reach the same conclusion as I do or not, at least we all agree that loudness normalization will level the playing field and permit dynamic material to coexist with squashed material and take away the pressure on artists to compress their material more than they desire it to.

Hope this makes my position a little clearer  :-)

Bob

Bob Katz - Loudness: War & Peace
Reply #6
Thanks for chiming in, Bob; a pleasure to have you aboard. 

Listening to those two versions of the country artist's master you made was a real eye-opener; even the second one you played back, which was the more open of the two, still sounded quite compressed/rather lifeless for my tastes.  It represents everything that is wrong with radio these days, in my mind, and it has certainly spread to consumer playback media as well.

This practice is a real shame, and the more people who know about it and stand up against it, the better the music community will be for it.  Thank you very much for putting this out.
FLAC -2 w/ lossyWAV 1.3.0i -q X -i

  • Bob Katz
  • [*]
Bob Katz - Loudness: War & Peace
Reply #7
Thanks for chiming in, Bob; a pleasure to have you aboard. 

Listening to those two versions of the country artist's master you made was a real eye-opener; even the second one you played back, which was the more open of the two, still sounded quite compressed/rather lifeless for my tastes.  It represents everything that is wrong with radio these days, in my mind, and it has certainly spread to consumer playback media as well.

This practice is a real shame, and the more people who know about it and stand up against it, the better the music community will be for it.  Thank you very much for putting this out.


Yeah, I mentioned that the more "open" version was "pretty competitive for 1999" but I agree it is far too compressed for my tastes. The irony is that it wasn't accepted by the "20-something" producer who wanted it to be slammed for 2011..... 

Fortunately, I still get plenty of work that allows me to produce great-sounding material, I just didn't demonstrate any of it in the video  :-).

  • GeSomeone
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #8
thanks for bringing this to my attention, and Bob for addressing the issue once again.
It ends rather optimistic, I hope it will happen, but won't hold my breath. Replaygain, as an practical implementation of such a "loudness normalisation" has been around for quite some time (I used it for some time even before foobar2000 existed, yeah the Musepack days). But the music producers are not aware of it.
The difficulty is that in order to go to "loudness normalised sound" you have to make the step to attenuate most music, there is no headroom to go up (we know when this is compensated by turning up the amp, all is well). I fear this will be hard to sell to the mentioned groups (iPodders, loud music producers, radio stations, etc.).

Fortunately some artists and producers care for their sound. It is such a shame when carefully recorded albums on hi-tech equipment end up sounding like (e.g.) Metallica.
In theory, there is no difference between theory and practice. In practice there is.

  • Wombat
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #9
Nice video, thanks. It is indeed a good wish to have some normalized sound in the future so that compressing makes no sense for the producer.

I just lately was frustrated again about getting good sound and how complicated it became. I have build my own speakers and fine-tuned them over time to sound good to my taste in my enviroment, in my room.
The biggest problem meanwhile are the recordings.
It is not like it sounds right if the speakers sound as neutral as possible. It isn´t that simple.

You can have even these compresssed recordings sounding pretty good when you muffle on purpose and create a layed back sounding speaker. Even many so called Hires distributed material is mastered HOT still. Just listen some B&W society of sound recordings. Of cause with most B&W speakers you need exactly this sound to make it alive  It may sound shabby on other speakers, that´s a dilemma.
The problem then is that other recordings, especialy of the kind there wasn´t much mastering at all sound damn lush and boring.

What i noticed is that this "Dynamic Range Meter" DM value more and more shows up in relation of judging the sound quality of recordings. This number alone tells nothing but you see the movement may already be there.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Bob Katz - Loudness: War & Peace
Reply #10
Interesting video. Thanks for posting the link.

Bob Katz - Loudness: War & Peace
Reply #11
It would seem that if the source material is going to be clipped/compressed to this extreme, there would be nearly no call for a 24 bit consumer audio format, yes?  High resolution clipping...?

Maybe it even argues that we should drop down to 8-bit recordings and save some space?
Was that a 1 or a 0?

Bob Katz - Loudness: War & Peace
Reply #12
Some modern recordings would justify a dithered 4 or 5-bit encoding.  If you can mask all bands of noise by a good 15-20 dB, that's pretty much all you need.

And, really, there isn't a call for a 24-bit consumer audio format, but let's not go down that road. 
FLAC -2 w/ lossyWAV 1.3.0i -q X -i

  • lvqcl
  • [*][*][*][*][*]
  • Developer
Bob Katz - Loudness: War & Peace
Reply #13
Maybe it even argues that we should drop down to 8-bit recordings and save some space?


http://www.hydrogenaudio.org/forums/index....st&p=499409

  • hellokeith
  • [*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #14
David & Bob,

I don't think it is fair to discuss the loudness issue as if it exists in a vacuum.  Are not the vast improvements in recording technology, especially microphones, part of the problem in the last decade? Mixers / producers have more sound to deal with..

  • shadowking
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #15
I think this is all going nowhere until we find out what is happening in the human brain - The young in particular are 'addicted' to a clipped louder sound. They are also addicted to a lot of visual stimuli, food (sugar, caffeine). I think it is a excitatory  / stress response  - Dopamine / nor epinephrine are elevated.


Bottom line except a minority everyone else is going for this loudness even if its eventually fatiguing.
wavpack -b350hhj0s0.7cc

  • Gecko
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #16
To my ears the final master seems to have less bass and more highs. Apart from that they sound quite similar: overcompressed and distorted. I can't hear more dynamic range in the first master. Are my ears too bad? Maybe I'm so used to this sound already.

I'm also wondering if the additional steps taken to punch up the volume, like stepping into the analog domain and clipping the converters, are necessary. Couldn't this also be achieved by throwing in something like Waves' L3 Ultramaximizer? Maybe that would be too transparent? Actually, I feel that the advent of Waves' look-ahead limiters was fuel to the loudness fire: easy to use and less audible side effects than what people were used to.

Overall I am really upset with this loudness race. It seems to reach new heights with every new CD that I buy and transcends into non-mainstream genres. I have a few CDs which could be labeled as medieval music. Very different from "in your face" rock music. But judging by my ears and replaygain, these are also being mastered increasingly hot over the past years. The results are sometimes absolutely horrible.

  • GeSomeone
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #17
[..]there would be nearly no call for a 24 bit consumer audio format, yes?  High resolution clipping...?

Almost the only reason for higher resolution formats would be the hope that they it would be mastered less agressive. If not ... useless.

To my ears the final master seems to have less bass and more highs. Apart from that they sound quite similar: overcompressed and distorted. I can't hear more dynamic range in the first master.

Overcompression often leads to less bass, typically in (heavy) rock music. Like the bass goes down when the drums kick in. Everything in the mix is fighting for the highest level.
I heard the biggest difference in the snare drum and the vocal.
In theory, there is no difference between theory and practice. In practice there is.

  • smok3
  • [*][*][*][*][*]
  • Moderator
Bob Katz - Loudness: War & Peace
Reply #18
<rant, slightly offtopic, but totally on for me>

Thanks, a kind remainder to keep out of the "intelligent" mixing suite with all the "smart" guys around there (I will just stick with video).

soundguy: "You know, our hard drives are so full, since we record everything to 128 bit, so that ..."

me (inside: "Why would you tell that to the person that handles HD uncompressed video?")

me: "Ever heard of FLAC before? ... No, actually the things you do, 16 kbps HE-AAC will be just enough and will give you plenty of room, did we solve that?
Now, can you move that text 2 seconds to the left before i smack you?"

</rant>

It is a true story (not just based on a true story).
  • Last Edit: 20 November, 2011, 10:50:25 AM by smok3
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

Bob Katz - Loudness: War & Peace
Reply #19
128-bit wordlength??  I didn't even know that existed yet...

Well, now I know that we're going to see (at least) a kilobit wordlength emerge within the next decade.  Now, pardon me while I have my cake and eat the world too.
  • Last Edit: 20 November, 2011, 11:48:54 AM by FreaqyFrequency
FLAC -2 w/ lossyWAV 1.3.0i -q X -i

  • cliveb
  • [*][*][*][*][*]
  • Developer
Bob Katz - Loudness: War & Peace
Reply #20
I fear that Bob Katz's expectation that loudness normalisation will help to reduce the extent of dynamic range compression is unfounded. As long as people's primary mode of listening remains via mobile devices in noisy environments, compression is a requirement. And this is a cultural issue - little to do with engineering.

Actually, come to think of it, we may be in a vicious circle here. Hypercompressed stuff sometimes sounds OK-ish on an iPod in the street, but sounds dreadful on a home stereo. This may actually encourage a shift in listening behaviour away from home systems towards mobile devices - which in turn puts pressure on making things ever more compressed - which turns people away from their home systems even more, etc, etc.

The correct solution is for playback devices to include compression facilities, so that it isn't necessary to trash the actual recording for everyone else. My Sansa Clip is Rockboxed and has a compressor, which I use when listening in the car or on a train. Even my old Rio Karma had an AGC that served the same purpose (indeed, in some sense it's a better solution for *really* noisy environments). Until the dominant playback devices (ie. iPods/iPhones) include a compressor, there is no hope.

  • Bob Katz
  • [*]
Bob Katz - Loudness: War & Peace
Reply #21
You're right. GOOD Compression facilities for mobile devices (including car) are needed. A wide range music recording cannot serve all venues at once. The technology is available, so it's only a matter of time and the will to do it.

BK
  • Last Edit: 21 November, 2011, 11:44:23 AM by db1989

  • krabapple
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #22
Offloading compression onto playback devices is really the only hope.  And there's no reason the industry couldn't adopt it.  It's akin to having the 'stretch' mode  for those who can't abide 'black bars' -- i.e., accurate aspect ratios -- on their TVs

  • knutinh
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #23
You're right. GOOD Compression facilities for mobile devices (including car) are needed. A wide range music recording cannot serve all venues at once. The technology is available, so it's only a matter of time and the will to do it.

BK

I believe that the gain applied by common compression algorithms is actually a quite smooth function over time, possibly extended into 3 or a few subbands. My point is that this kind of information can easily be embedded into the distributed file as metadata in a losslessly compressed form. The advantages over standardized playback-side regular compressors and/or studio-side compression would be that:
1. The raw pcm data could be the sound engineers/musicians vision of how the recording would sound best
2. Any playback device could apply the over-compressed gain-function at minimal cost
3. Proprietary algorithms could be used for calculating the gain as a function of time - i.e. large freedom for future compression trends, and the actual algorithm for calculating the gain would be concealed
4. Very low quality devices (cell-phones on loudspeaker mode) could evaluate the before/after mix and apply their own algorithms to fit as much as possible within a limited envelope

I actually think that this is the kind of stuff that will decide if produced music will survive as a commercially and artistically viable art-form, or only shelf-ware background noise for doing something else. Sadly, the business have wasted a decade or two on "hirez" music, copy protection and other schemes that was not in the customers interest.

-k
  • Last Edit: 21 November, 2011, 02:17:01 PM by knutinh

  • googlebot
  • [*][*][*][*][*]
Bob Katz - Loudness: War & Peace
Reply #24
Meta-data which lets the producer somewhat control the amount and type of client side compression would be the way to go. Something like Dolby Digital's dialnorm complemented with a compression type info field and maybe better loudness estimation (e. g. R128).