I'm there too.
I find DR14 perfect! I hope this becomes finally a standard. I have always been putting my two cents in my environment toward this.
For those thinking that it's limiting, think that the idea to avoid the loudness war, is to stick to a fixed perceived loudness, so that nobody sounds louder than anything else. If that's not respected, Jose may think that -13.5 is just there, and then John why not -13.2 if Jose did -13.5, the Marah comes with -13.0 that's onle 0.2 dB above... and four years later we are at -4  . You get it, that's what happened already and what's to be avoided.
DR14 is not a sacrifice, it is very reasonable. I have been using -12 for CD and -20 for 24 bits for years. There is plenty of SNR below and enough headroom above. The cinema industry strictly follows the -20 rule since many years ago. It is a need.
 
Your latest song can sound louder even without compression? And what's the problem in leaving some headrom unused? When you later do a thing that requires more/quicker dynamics, you'll be happy not having to overcompress it to sound as loud.
I also have been using ReplayGain for all the music I hear, at home, in the car, or with my RockBoxed iPod. My librarian, MediaMonkey, analizes the perceived loudness of the track and record separately, and saves the values in tags. All the players I use automatically adjust the playback gain of each song according to these ReplayGain tags. Overcompressed stuff obviously keeps sounding as crap, but not any louder than anything else. It is a big pleasure to hear music without having to adjust the volume for each song or album.
 
Now, the first problem I found: I don't find any technical description of the perceived loudness model they are proposing. Just the RMS level is not enough, a more complex frequency/level weighting is required. I know because prior to ReplayGain I tried to standarize and found that same RMS levels can be perceived as very different loudness depending on the spectrum. But ReplayGain, a public open source proposal that is getting good acceptance, nailed it perfectly. I really hope that this TT-DR meter actually applies some adequate weighting... I just downloaded it and I'll check it tomorrow.
 
The second problem. They are just starting and now the meter download is not free. I had to pay $30 for membership in advance, without being able to access any relevant technical information about it. I won't regret if this ends in a success, but I believe this is not the best way to achieve a  massive acceptance, and that's actually what's needed to be a success.
 
Despite of these little annoyances, I'm going to heartly support any initiative in this direction.
 
 

bitflipper
I have read that its calculation is based on the loudest 20% of the song. The theory is that this will eliminate fade-ins, fade-outs and full rests from skewing the numbers.

We need to know the technical details. And DR needs to be good enough as evaluating the perceived loudness. I knew about it today and I'll check it tomorrow.
ReplayGain is a proven algorithm to evaluate overall loudness weighting different frequency bands differently, etc. and I can attest that it works perfectly. If DR is any worse they should consider adopting ReplayGain or taking it as an idea, it is a public open proposal that already has meaningful acceptance. 

Unfortunately, it also means that if you have a song that intentionally super-compresses the big finale, the song will rank low on the DR scale, even though the song might be quite dynamic overall.

My own stuff sounds most satisfying with a DR rating of 10 or 11, which, according to the Pleasure Police, is (moderately) overcompressed. But it's an artistic judgment, and I'm finding that attempting to establish a numeric target for all music is probably a losing proposition.

Some of my favorite music (both my own and commercial recordings) comes in at 14. But other favorites come in at 8. I wouldn't want to see a standard in which music exceeding some "optimal" DR level would require a warning sticker.

What's the problem in just lovering the gain those 3 or 4 dBs and leaving those unused?
A loss of 4 dBs of SNR leaving your average "only at 98 dB" above floor will compromise your "artistic judgement"?
 
Please consider this, Dave, you are a very knowlegable person and this deserves a more serious consideration.
I don't doubt you understand perfectly the advantages of adopting an standard.
You say most of your stuff measured 10 or 11. I have been trying to stick to 12 for a long time, I'll see tomorrow how it compares with DR rating. 14 is very reasonable spot to provide a large SNR below and good enough headroom for most stuff.
We only must wish that the algorithm is right evaluating perceived loudness, and that a standard is massively accepted.
 

ReplayGain is a proven algorithm to evaluate overall loudness weighting different frequency bands differently, etc. and I can attest that it works perfectly

And therein is my whole problem with the Pleasurize group. ReplayGain has been around for years, it works well, and if your playback device supports it makes large volume swings between songs a thing of the past. It's like dialnorm for music. Why on earth did Pleasurize feel the need to reinvent a system? At the very least they could've added suport for the latest ITU standard for loudness measurement to a replaygain system. That would've been useful. Plus there's already the loudness/dynamic range meter from TCElectronic. We don't need yet another system to fragment the various loudness systems.  I don't think I'll be supporting them.

What's the problem in just lovering the gain those 3 or 4 dBs and leaving those unused? A loss of 4 dBs of SNR leaving your average "only at 98 dB" above floor will compromise your "artistic judgement"?

It's because we're not talking about peak levels and not about SNR.

Sure, if I find my master bus is too hot, I just turn every track down. Nobody will notice if it's -3db versus -0.1db like your typical commercial recording. It's also not just about preventing clipping, but also about giving your bus limiter something to work with.

Turning your peak levels down by 4db is not the same as lowering your average RMS by 4db. One is a technical decision that listeners won't even notice; the other is an artistic decision that dramatically affects the way the song is perceived.

Some people might actually prefer the sound of DR4 over DR14. If your target market is diehard Metallica fans, for example, you'd better get that compression ratio up there or you won't snag them. It's not about being competitive loudness-wise, but simply meeting expectations.

My personal tastes are all over the map. One of my favorite recordings is Alisson Krauss and Union Station's "New Favorite", with an average RMS of -14db and a DR rating of 11. It's mellow progressive bluegrass, great falling-asleep music. And it's DR11.

For wake-up music, a piece I like is Jordan Rudess' version of Tarkus. It's much more agressive than ELP's original: -9db average RMS, DR7. It does not sound overcompressed to me, just very aggressive and loud. Definitely not for falling asleep to.

For pure uncompromised quality, there are few better pop examples than Paul Simon's "Graceland". -18db average RMS, DR13. (Note the discrepancy between the DR rating and the average RMS, a symptom, I believe, of the DR meter looking at only the loudest parts of the song.)

Stevie Wonder is often held up as a shining example of dynamic integrity. "Yester-Me, Yester-You" has an average RMS of -13db, a DR rating of 11.

My point here is there is no magic DR number where everything starts to sound great. Even a DR rating of 10 or 11 isn't necessarily dynamically-compromised.

EDIT:
I dug through my record collection looking for a super-squashed song that I nonetheless enjoy listening to. There aren't many, but I came up with "Jesus of Suburbia" by Green Day. I consider it a pop masterpiece, even if I wouldn't listen to it twice in a row. -6db average RMS, peak amplitude 0db, and it gets a DR rating of 5. Would it be a better song at DR7? DR10? DR14? Can't say. But I do think that people who buy Green Day records don't expect subtlety, and like it or not, if I were GD's producer I'd recommend heavy compression because that's their niche and their fans' expectation.

Dave,
What I meant is that if you lower the overall level after all the processing, only the level of the output changes, nothing else, not level ratios nor anything. I thought it was obvious. You choose the dynamics and everyting. You may compress as much as you want.
 
Rosabelle,
I agree with you that ReplayGain fixes the issue nicely. If we could assume every player used it, nothing else would be necessary.
But we can not assume that, at least for now, and music keeps been distributed highly overcompressed. Setting a standard for perceived level, like the cinema industry uses, is a good thing anyway. It is not incompatible with ReplayGain, and BOTH can help to solve or mitigate the overcompression plague. The problem is there, is not going to dissapear easily, and we should not disregard any measure that helps.

After testing a bit the DR meter, I'm higly dissapointed. It only uses RMS value averaging, and this is far from accurate regarding the perceived level. The simple documentation has many errors and misconceptions.
At this point I agree that the DR initiative is not very serious.

I wished DR was better. But is still better than nothing, and I'll try to help them improve the measurement algorithm. There is nothing to invent, the ReplayGain is proven to work very well and is open and free, non propietry. I hope they will understand RMS is not good enough for loudness evaluation.

Dave, What I meant is that if you lower the overall level after all the processing, only the level of the output changes, nothing else, not level ratios nor anything. I thought it was obvious. You choose the dynamics and everyting. You may compress as much as you want.

I'm confused, so I guess your meaning was not obvious, at least to me.

I agree, leaving some headroom to avoid intersample clipping is not a problem. No one is going to notice if your final product peaks at -3db versus -0.1db. But that won't affect your DR rating, since the DR tool (AFAIK) does not care what your peak levels are, only the dynamic range.

I was pretty jazzed about the DR Tool in the beginning, but it's turned out not to be as useful as I'd hoped it would be. Whenever it goes into the red Forbidden Zone, it's merely affirming what I already know.

bitflipper
But that won't affect your DR rating, since the DR tool (AFAIK) does not care what your peak levels are, only the dynamic range.

You're right, Dave, I apologize for the confussion.
DR only takes into account the peak/RMS relationship, regardless of level.
LOL, I guess this is worse after any new detail I discover. I would support a standarization of perceived loudness level, not forcibly having to respect a given peak/RMS ratio.
 

I first consider the performers concept/presentation of a song.  Then we consider the end result.  At that point we consider the song and "what it needs".  If it is a driving song, I might add a LOT of compression or even SONAR's Boost11.  Some C&W songs are like that.  If it is a quiet, soulful song, I may not use compression at all.  Most time, I do use some compression, but v-e-r-y sparing in it's action. It makes the end result more listenable.

i'm bumping this....
because.....

hm, well, i don't know, i'm curious to know where folks who were into this conversation back uin 09, are sitting with this now.

any changes?

Firstly it is no different to using the K system approach. All you need is a VU meter plugin and you can calibrate that meter for -14 as the ref and it will show full scale or 0dB VU when the average rms level is around -14.

The K-12 is good level too but in reality none of this means anything unless every piece of music produced from now on adheres to a system like this and as someone who produces music professionally I can honestly say that many clients are still wanting their final mastered tracks to be loud and as loud as the next guy.

Good point bats about what is happening now. Well what is happening now is we have some fantastic limiters around that were not even around in 2009 and the good news is they seem to be able to make the average rms level higher without smashing the mix and somehow they seem to keep some dynamic range in there at least. (PSP Xenon)

I agree though. I work at K-14 for everything mostly and end up with a mix and mastered tracks also at K-14 and I would love to be able to just stop there. K-20 is a fantastic ref level too so why not limit it to that as well. That sounds even better. I often start projects at K-20 and move it up to K-14 and then ultimately a higher level for the client.

Maybe it will happen. They are starting to set standards for rms levels and volume levels for broadcast TV sound now so the adverts are no louder than the program material so that is a start. Who knows maybe some of that will flow onto audio production too. And we have also had standards for film sound for years where K-20 is the ref level and the average (dialogue) SPL's out front are 85 dB SPL at K-20. Meaning that things can get as high as 105 dB SPL if needed and as we know they do use it, often!