Average Music Transients/Peaks Data across genres of Music ?

[krivium]

I thought LKFS is the same as LUFS? Reading about them they both seem to implement K-Weighting? The weighting thing seems very relevant, but when reading about LKFS/LUFS it is hard to find any mention or examples on what kind of target curves are defined, still seems they are looking mostly at VU levels?

I've done a bit of research about Lukfs and Lufs because i wasn't confortable about it.
So even if both use the same math and are interchangeable they differ in their results in how it is displayed because one is 'K- weighted' and is loudness referenced to full scale ( LUKFS, in America so ATSC), the other isn't K-weighted* and is referefenced to an absolute unit so LUFS ( in EU so EBU).

In practice it doesn't really change their behavior ( thanks god!) and are interchangeable.

The target volume curve doesn't exist in music because there is still no global standard used/defined.
But if you know your music will be played through Itunes or Youtube or,... they defacto impose a leveling.... So it makes sense to follow this as a target: for Itune this is around -16/-16,5 dblu. And it seems others are not far away so i would follow this recomandation.

And guess what, it is the K-14 dynamic range with 2db headroom allowed which is nice as it gives you an equivalent to True Peak margin embeded, so no more intersample peak allowed ( bye bye cold harsh digital sound as this was the main reason for the 'coldness' digital was blamed for) and typical pop music level!

This mean in case you use K-14 for your monitoring you can safely use any regular pop song produced before loudness war as reference.

* the K weighted curve is interesting. It follows what i told earlyer about compression and low mid high... in fact they implement a 12db highpass at 100hz and a shelf boost of +4db from 2khz and up. The reason being we ( our brain) integrate loudness from 100hz and the boost is to mimic what our head implies as acoustic change.

 

[mondogenerator]

Well most of that was over my head, and I'm not an expert mastering operator, so I still fumble a little with compression WRT frequency bands. Usually I muddle through ok.

As far as mastering levels....I don't know what is up with recording engineers...

I just set the waveform Crest to -12 to -15 dBFS and be done with it.

 

[krivium]

I just set the waveform Crest to -12 to -15 dBFS and be done with it.

Hi, this is low. The reference level we talked about are for rms level not crest/peak.
If your music is 'typical' wrt DR -15dbfs is not bad target for rms level of the loudest part of your loudest song (in case of an album).
That is why -16dblufs target is even better as it allow for 1/2 db headroom to stay away from peak intersample distortion.

If you set your max crest/peak to -15dbfs you 'waste' around 10/12 db of potential dynamic usable range ( it is fine as you won't clip your output but you'll loose in accuracy wrt dynamic: you waste 'bits' of resolution).

Iow, if you want to stick to ppm ( peaks/ crest) make them bewteen -3/-6dbfs.

The whole point about K-metering is to not use PPM anymore when working digital ( as it doesn't represent anything representative of our perception of level but is a 'technical' parameter indicating the 'limits' of ADC or digital chain) but allow to work with VU meter instead (which are much more representative of how we perceive loudness - integration time of 300ms is more or less the same as our brain integrate level).
Working with VU was the way it was done for decades with analog gear and for good reasons... When digital arrived we switched to PPM because of the gear limitation ( 16bits 1st gen converters needed to be the closest to 0dbfs) and if you ad to this the 'loudness war' this led us to the situation we had seen.
Hopefully this will change in the next years ( i hope).

 

[mondogenerator]

Krivium,

I agree with your points, it is low, and it does "waste" bits...

Perhaps 4 of 24bits, and it is a very conservative set up.

It is my opinion, that I can record a natural, good track with 16 bits: 24 bits minus 4 LSBits at noise floor, another 4MSBits at max dynamic range.

The point I was making is that I will never record too close to 0dBFS, -6 to -9dB is perhaps better.

Even 3dB headroom is better than a bad engineer sitting at -0.5dB and relying on limiters to stop FS clipping (bad engineer = me)

I don't use limiters, they're one type of compression I cant stand the sound of, when used badly, or sometimes even when used well, the waveform still ends up distorted.

Even using excessive "headroom" in my simplistic way, recording in 24bit, then amplifying back to -1dB at mastering stage, I have no real noise issues that cant be resolved by gating/noise reduction SW.

After all we are talking Mastering, rather than channel recording/dubbing (I.e. if all my channels are -9dB when it comes to master, but are clean, noise free, then I end mastering back up closer to 0dBFS but never above about -3dB, and all is very well indeed)

In the past I had a penchant for recording my drum track, no comp, no limiter, full DR - then afterwards using a light compressor (fairly light, maybe 2:1)

That seemed to work well for me, but I'm not mastering for Floyd, or Dire Straits!

(Sultans of swing has the most lovely DR, and warm yet not too dark ride cymbal presence)

 

[krivium]

The point I was making is that I will never record too close to 0dBFS, -6 to -9dB is perhaps better.

Yes this is my point too: with 24bits ADC you don't need to use 'practice of the past' anymore ( as close to 0dbfs). Headroom is very important when recording digital and as long as you don't clip at input you can always make up gain elsewhere in the chain.

Even 3dB headroom is better than a bad engineer sitting at -0.5dB and relying on limiters to stop FS clipping (bad engineer = me)

I don't use limiters, they're one type of compression I cant stand the sound of, when used badly, or sometimes even when used well, the waveform still ends up distorted.

I agree with you, automated level treatment should be used with caution and not be relyed on only to keep you in safety range: the way you arrange your gain structure is the main way on which you have to rely, then you can use a limiter in case of and with intelligent parameter set up.
Iow limiter on mix buss shouldn't give you more than 2db reduction imho.
Of course there is no rules so you can make it much more from them but there is a price to pay. The most important is that it fit your track ( genre, the goal you have).
The example of 'shredding' i've linked in a previous link is what this track and his kind of production was asking for. In the same album there is much more soft travks which are processed very differently as the tracks main instruments are bass with very ambient deep pad and this technique would destroy the space needed by the song...

Even using excessive "headroom" in my simplistic way, recording in 24bit, then amplifying back to -1dB at mastering stage, I have no real noise issues that cant be resolved by gating/noise reduction SW.

Thanks to 24bit we don't have this much concerns about noise in the digital domain but your analog front end may induce excessive noise, so theorically it is still better not to allow too much headroom neither. All this is theorical though so in practice if you have found a way that work for yourself that is fine.

After all we are talking Mastering, rather than channel recording/dubbing (I.e. if all my channels are -9dB when it comes to master, but are clean, noise free, then I end mastering back up closer to 0dBFS but never above about -3dB, and all is very well indeed)

I don't really make a difference with tracking or mastering or anything else. This is all a question of headroom allowed for occasional transient to pass without being clipped.
Transient have to be as intact as possible as there is so many information about the source in them: the family of instrument, the size of it,.... its all there in the first milliseconds.

But if your track needs heavily distorded sound, being digital or analog go with it!

In the past I had a penchant for recording my drum track, no comp, no limiter, full DR - then afterwards using a light compressor (fairly light, maybe 2:1)

That seemed to work well for me, but I'm not mastering for Floyd, or Dire Straits!

(Sultans of swing has the most lovely DR, and warm yet not too dark ride cymbal presence)

Who cares about the way you do thing as long as it suits you and you don't make technical aberation?
I mean there is no rules as long as it suit your goal. But for certain kind of sound there is 'best practice'. There are no rules about producing and artistic choice, but once a path is choosen you have to know what it imply technically.

 

[KaffiMann]

I have not dropped the topic, been spending quite a bit of time figuring out stuff IRL.
Currently looking into this whole music mastering thing across a number of different DAW's, and what really annoys me is that they are all so different in every way. Not just talking about work flow or layout.
What they all seem to have in common is: they have some kind of meter to watch the output of channels and busses eventually routed to "Master". But that's where it all stops. Between the various DAW's the levels and scales are all over the place. Seems very hard to accurately compare them without writing specific pieces of music for each and every software, just importing WAV tracks and adjusting levels can probably get you in the same ballpark, but using some compressors and reverb on the various channels can in some places light your db meters so far into the red that I have actually experienced clipping (on Master output), while the actual output is -18db.
Then with my newfound knowledge I went back to some other DAW's I'm more familiar with, and now I find myself not trusting the various output meters at all anymore.

A db is not always a db, and a scale defined by clearly stated standards is obviously still open for some interpretation...

At any rate, still looking for some kind of DAW I am actually comfortable working with, this may require quite a bit of time and effort...

 

[Hiten]

At the risk of stretching my limits here... This lack of range is why my music collection ended someplace in the early 1990s. Everything was getting louder and most of it was actually giving me headaches so, at some point a few years ago I just stopped collecting it.

I've been reading up on things like the "Loudness War" and the way modern music is mixed and mastered. In my opinion the big problem is that many modern engineers have missed the boat when it comes to the way you mix digital music.

They've continued shooting for the 0 db mark as an average, as you would on an analog console. But in digital formats 0db is where clipping starts, it's a hard limit, there's no "overhead" so they can't just crank it up like they used to.

If we look at the standard consumer audio levels, the equivalent voltage output (1v peak to peak) actually occurs on CDs and DACs at about -15db... and that's where they should have been mixing.

This video might provide a better insight... YouTube (it's a bit long, but worth seeing)

Very sorry for late reply. Will go through it. thanks to everyone. Dont know how I missed the thread. Lots of things to learn. And with limited technical knowledge I take time to understand.
Regards.