true peak and what to do

thanks for the new True Peak implementation! very timely especially in times of over compressed and limited mixes to detect inter samples peaks.
but how do i deal with that information?
looking at a final CD master image file set to -0.3bd headroom. after analyzing the file, WL tells me that i have a max peak of -0.3db but a true peak of +0.7db. do i need to ideally lower the volume of the whole master image file by 0.7db to stay out of trouble?
any experience?

thanks

Yes. In the EBU recoomendation, then even recommend -1 dB True Peaks level.

I recall that the new Loudness Meta Normalizer can do this for you, eg.:
2013-05-24_07-39-35.png

thanks phillipe.
that is helpful.

kamal

The recommended minus 1dB TPL headroom is true for R128 relevant content. This means everything which goes officially to Broadcasters like commercials and film mixes etc.
The music industry is not aware about a sound and distortion problem at all.
Therefore I recommend to use target peak values between 0 and minus 0.5 (TPL).
Minus 0.5 TPL is ideal because this is the theoretical maximum underread of the TPL algorithm.
On the other hand you do not loose too much of level in order to play the loudness game
on a solid level :wink:
Minus 1 TPL headroom is definitely too much for commercial CD releases today.
For those of you being interested in an upcoming mastering book including WL8 workflow recommendations may pre-order and support the book finalization: finalization and print of THE AUDIO MASTERING WORKBOOK | Indiegogo
Best regards,
Friedemann Tischmeyer

Pity because there is a big problem.

Could it not be argued that all the hyper-compressed mixes being produced by said industry are inherently damaged? Not everyone in the industry subscribes to the same thing, and this is part of the problem. There is loudness anarchy and peak confusion.

The whole audio industry is slowly and painfully moving towards a loudness based paradigm, as opposed to a peak based paradigm. When the whole broadcast industry respects R128, or something similar, from delivery to actual broadcast, hyper-compressed mixing and mastering will cease to have any real value, and will actually sound rather weak - all of which I am sure you are aware.

Is it therefore not wise for the music industry to start becoming aware of how damaged all the hyper-compressed stuff really is? and to become aware that they are slowly but surely being left behind?

IMO only if you adhere to a peak-based maximisation, and hyper-compressed, paradigm. There is really no audible improvement in consistently pushing peaks to between 0 and -0.5 dBTP as you recommend, at least certainly not as a blanket recommendation for all types of music programme.

Would you not agree that you might do far better by concentrating on the loudness and dynamic range of the audio rather than the peaks? The ear responds to loudness and dynamic range far more than peaks, so taking care of these elements is probably far more important than peak maximisation. You could let the odd peak hit -0.5 dBTP, but -1dBTP seems a reasonably good target for the majority of the time.

Surely, the EBU recommendation (-1dBTP) is chosen from the point of view of protection ie: ensuring that inter-sample peak distortion is avoided in the reproduction chain, or if the master was to be converted to a lossy format at a later date. It is not some kind of manner in which we can apply a peak-based paradigm to a loudness meter. Peak level settings as applied when using peak meters and peak-based mixing habits ceases to have the same relevance in the world of loudness meters. Clearly, an EBU R-128 meter is firmly concentrated upon the loudness values, and in this setting the peak values become secondary, and somewhat arbitrary, (although never to be completely ignored). Where you set your loudness values and how much compression you use are of course different discussions entirely.

All just my humble opinion, of course.

Not necessarily the whole master. It depends how many peaks you have at that level. You might have only a single peak that hits 0.7dBTP. As an alternative to PG’s advice, you could try inserting the brickwall limiter at -1dBTP to overcome the issue, and be sure of avoiding inter-sample peak distortion.

to clarify:
when i say “whole master” i am looking at the final wav file of the CD after rendering to 16b44k. those TP’s are few only and i was inquiring about best practice to deal with them to avoid any problems in playback of the manufactured CD.

and i am totally in favor of mixes and masters with high dynamic range (generally K-14 as guideline, have not explored the EBU R-128 standard). but practically every master will eventually have to pass a limiting compressor (i’m using the UAD Limiter) set to -0.3db which will still lead to the occasional TP in plus values as described.
is there a limiter that tracks True Peak?

PS: can anyone recommend easy reading regarding the EBU R-128 standard and implementation?

IMO only if you adhere to a peak-based maximisation, and hyper-compressed, paradigm. There is really no audible improvement in consistently pushing peaks to between 0 and -0.5 dBTP as you recommend, at least certainly not as a blanket recommendation for all types of music programme.

It reduces the risk of ADDITIONAL subsequent distortion (within D/As and encoding processes)

Would you not agree that you might do far better by concentrating on the loudness and dynamic range of the audio rather than the peaks? The ear responds to loudness and dynamic range far more than peaks, so taking care of these elements is probably far more important than peak maximisation. You could let the odd peak hit -0.5 dBTP, but -1dBTP seems a reasonably good target for the majority of the time.

This would be ideal but we live in this real world and this means that we have to deal with the current situation and this is LOUDNESS WAR :slight_smile:
Seriously you would loose 99% of your clients as mastering engineer if you would go for the ideal approach.

Surely, the EBU recommendation (-1dBTP) is chosen from the point of view of protection ie: ensuring that inter-sample peak distortion is avoided in the reproduction chain, or if the master was to be converted to a lossy format at a later date. > It is not some kind of manner in which we can apply a peak-based paradigm to a loudness meter. > Peak level settings as applied when using peak meters and peak-based mixing habits ceases to have the same relevance in the world of loudness meters. Clearly, an EBU R-128 meter is firmly concentrated upon the loudness values, and in this setting the peak values become secondary, and somewhat arbitrary, (although never to be completely ignored). Where you set your loudness values and how much compression you use are of course different discussions entirely.
[/quote]

Once again. This is the idea and basic concept of R128 and this is the future. But this will never ever work for CD releases.

I am happy to be not alone as a dynamic warrior

is there a limiter that tracks True Peak?

Yes, included in WaveLab:
2013-05-25_11-17-04.png

Yes, as PG says, use the supplied brickwall limiter set to -1dBTP with detect intersample clipping active. This ensures no true peaks above -1dBTP. Or set the true peak according to Friedemann’s recommendations above.

Yes, agreed, R128 as a whole may not be applicable to CD releases but only in the current hyper-maximised scenario… there are a number of CD releases whose tracks fall close to or within its limits (admittedly many of these were produced before the loudness wars). However, in the case of dealing with true peaks, which is what I was referring to here, there is no real reason why all CD releases should not adhere to -1dBTP (and this not suggesting that we should follow R-128 recommendations, but that where peaks are concerned -1dBTP is probably a good guideline anyway). In any case, I think most would agree that some degree of headroom for peaks is good practice in CD mastering. Just how much headroom would seem to be debateable.

About the EBU R-128, there are 2 sides:

  1. The loudness and true peak measurement part.
  2. The recommended values for these measurement, for the broadcasting needs.

#2 is obviously not useful for CD Mastering.
#1 is absolutely useful for any audio production, incl. CD Mastering. Because the algorithms proposed by the EBU R-128 are universal, ie. independent of any broadcasting context.

the water is pretty muddy in music because many modern converters have headroom above 0 dBFS to reproduce intersample peaks. IOW distortion is not a given, only implied. With tools like True Peak it’s now possible to avoid distortion on any system, even for crappy older converters like those found in many car stereos. So that’s a good thing.

I’d take issue with Mastering Academy’s claim that -1 dBFS (or any arbitrary value that eliminates intersample peaks in a particular program) won’t fly in CD mastering due to loudness-envy. First I have seen a lot less pressure from customers, and with so many vinyl releases I expect the demand for distorted masters will continue to decline. More significantly, a good masterer can deliver a perceptually louder program than the hacks (read: mixer dude with L3 or Elephant), with peaks that don’t distort. IOW, I frequently deliver masters peaking between -.5 and -1 dBFS that sound louder out of speakers than the competition’s 0dBFS masters. Loudness is a function of average level and accumulation/integration time on meters, not peaks. There are all kinds of tools to enhance transients if you can’t figure out how to set up attacks on your dynamics chain, so it’s much easier to manage today.

The bottom line is that setting your peak level at -1dBFS is not necessarily going to give you a “quieter” master, nor will it necessarily eliminate intersample peaks (frequently overshoots are +3-6 dB!). True Peak and Meta Normalizer are terrific tools to see what’s really going on. Thanks PG!

So the cheapest car stereos could be an example of a worst case DAC? Something with no headroom? I’m actually looking for something like that to hear the effect. Maybe the cheapest portable player? I’ve heard the DAC in the iPod is pretty respectable (and I would assume has some headroom), so I would think I’m looking for something quite a bit worse than that. If anyone knows of anything that definitely has no headroom…

The cheapest dacs that would have problems would probably be off-brand portable players from the mid-90s. Dacs are a lot better these days and software masks many sins.

Current iPod lineups have decent dacs as you suggest, but the old white “gum-stick” iPod shuffle’s have some of the worst dacs and analog circuitry I’ve ever heard… noticably different than current iPods and iPhones (most iPhones are pretty good).

[quote=“Daved”] the old white “gum-stick” iPod shuffle’s have some of the worst dacs and analog circuitry I’ve ever heard… quote]
Thanks Daved. I’ll give one of those a try. I do wonder though how much current gear actually has built-in headroom to handle the worst cases. I don’t know where to look to find such information.

You can only test. :wink:

I’d say this: CDs have been pushed for over a decade. MP3s are polluted by the damage, but reconstruction is as much a software issue as hardware, so an intelligent transcoding algorithm can resolve this, and I bet it’s far less of an issue than it once was; once files are transcoded to 32 bit float space, the peaks can be very easily scaled into range without artifacts.

Of course actual clipping is still clipping. Talking about a single missing peak sample between a pair of FS values located 1/44,100 sec apart is fairly trivial in most band-limited programs, but as soon as you have 3 consecutive 0 dBFS samples in a row, you’re missing relevant data about both peak and slope which cannot be recovered. In these cases no hardware or software can restore your program to it’s original sound, and that’s what others here are referring to above.

I read about true peek only some weeks ago and i look to see Oscilloscope videos from Line Out or speaker or headphone out about it. because i do not believe it. But i find none. when rectangle signal in testtone generator is used the true peek Meters show +3.7 db or more with 14.4 khz testtone. but i want see what happen and i connect my 10 MHZ oscilloscope to headphone out, and there can not see higher peak and no clipping in the analog part. more text explain in the video text true peak Level Meters are wrong confirm with 10 MHZ Oscilloscope in this Video(no Videosound) - YouTube

Hey folks! I’ve been following this thread and there is really good info here. May I offer a thought?

I have 20 years of mastering experience and when I am working on an album or single and I want to bring the overall loudess/density up, I focus not on the limiter to start that process, but careful manipulation of creating that fullness from within the track, not from the outside, such as limiting. I also seldom use broadband compression either but rather I rely on processes such as M or S compression, carefully placed multi band compression, careful EQ choices too.

All if these things can bring up the impact and density of the track and yet still allowing it to breathe well and all WITHOUT creating additional peak signals. Processing in this way gives me a track that sounds incredible at any volume and when placed at the recommended levels for streaming be it -14 or -16 keeps my peak information very easy to manage. Just a thought…