KVR Audio

annode · Post by **annode** » Mon Dec 11, 2017 6:39 pm

Thinking out loud;

Loudness has two components, Peaks and Average.Either one alone can be made loud.
Add sounds together and it will get louder. Increase the gain of a sound and of course it gets louder.

Questions;
- What is happening when compressed music is perceived as unpleasant? ( I know what compression is doing to sound, but why does it become unpleasant at some point?)
...maybe when the number of sounds that happen together are increasingly pushed to the same level, then the sum is not only the total sound but also added inter-modulation of these sounds as they interact, in the form of discordant harmonics and distortion. How mush does the process itself add to this unpleasantness? How much of this is bio-mechanical within the ear itself or brain ?

- What process(es) and equipment are being used to set broadcast material to a target loudness level?
...if limiting, isn't the material still perceived as loud? Or maybe it's mostly peak limiting, removing one element of loudness. How do the engineers handle this and what's the thought behind it?

annode · Post by **annode** » Wed Dec 13, 2017 4:33 pm

annode wrote:Thinking out loud;

Loudness has two components, Peaks and Average.Either one alone can be made loud.
Add sounds together and it will get louder. Increase the gain of a sound and of course it gets louder.

Questions;
- What is happening when compressed music is perceived as unpleasant? ( I know what compression is doing to sound, but why does it become unpleasant at some point?)
...maybe when the number of sounds that happen together are increasingly pushed to the same level, then the sum is not only the total sound but also added inter-modulation of these sounds as they interact, in the form of discordant harmonics and distortion. How mush does the process itself add to this unpleasantness? How much of this is bio-mechanical within the ear itself or brain ?

- What process(es) and equipment are being used to set broadcast material to a target loudness level?
...if limiting, isn't the material still perceived as loud? Or maybe it's mostly peak limiting, removing one element of loudness. How do the engineers handle this and what's the thought behind it?

After writing this I went to Wikipedia for answers.
There was a theory concerning our brains perception of loudness which was interesting. It also had answers concerning the software being used and how it works.

Burillo · Post by **Burillo** » Wed Dec 13, 2017 5:10 pm

no need to go that deep. it's actually pretty simple: it's fatigue. we like dynamics, we don't like lack of dynamics. heavily compressed sound lacks dynamics, hence we don't like it.

Compyfox · Post by **Compyfox** » Wed Dec 13, 2017 5:16 pm

Sorry, currently busy with other projects.

Just real quick

both "limiting processes" are done with gain -> limiter ceiling, while a specific meter checks if you're basically "In the green".
Loudness Normalization uses the ITU-R BS.1770-x specs and a lower reference level (if you can see my signature, I've written a couple of posts how to do this)
"Peak Limiting" in recent years was a gut instinct - push as loud as possible, check with a RMS realtime meter ever so often if you're really "the loudest compared to the competition", see that nothing exceeds a certain ceiling (the infamous -0,2dBFS), render out and hope that it sounds great in MP3 form

"Peak Limiting" was the reason the K-System (v1) was proposed, and this also sparked the now infamous DR-Meter concept. Then a couple of years later, ITU-R BS.1770-x surfaced in the form of the EBU R-128 "preset".

People then freaked out - because it was advertised as "defacto 'standard' - everyone has to adhere to it", people thought this was for "mixing and mastering" (spoiler alert: it's only for mastering). This sparked many discussions, spawned the creation of various new metering tools (PLR Meter comes to mind) and ways how to "bend/break the rules" yet again (see other forums on the web), until some clever audio engineers finally sat down and said "let's do a compromise, and focus more on educating people" (resulting in the -16LUFS to -14LUFS range for regular releases).

That is massively simplified and scratching the surface of course, but it should give you a small peek at what happened in the last 10-15 years.

Daru925 · Post by **Daru925** » Sat Dec 23, 2017 7:28 pm

Hello i'm new to this and trying to figure out how to apply those recommendations.
So until there is a definitive accepted standard, (which might never be the case), what are some good targets?
From what i read/gathered, it seems some good compromises for now were:

-16lufs for streaming (mixing at -23)
-9lufs for cd (wav 44.1 khz 16bit)
what about mp3? Are lufs applicable to the jpeg of music ?

Mushy Mushy · Post by **Mushy Mushy** » Sat Dec 23, 2017 7:38 pm

99 lufs here.

Or is that just my balloons?? Confused...

Compyfox · Post by **Compyfox** » Sat Dec 23, 2017 10:32 pm

Daru925 wrote:From what i read/gathered, it seems some good compromises for now were:

-16lufs for streaming (mixing at -23)
-9lufs for cd (wav 44.1 khz 16bit)
what about mp3? Are lufs applicable to the jpeg of music ?

-9LUFS is too much. At this loudness you still need a STRICT -1dBTP max value (if not even -1,5dBTP) to be save from clipping.

If you really want to push for CD, a compromise is -14LUFS Integrated (considering that Momentary/MLk can still shoot up to +9LU higher for forte fortissimo passages, which could result -5LUFS MLk max!). But get used to the -16LUFS value, as this is considered an "the best of all worlds" scenario.

Don't listen to the nonsense that "club music has to be loud" or anything similar along those lines. Lower values are better for your transients.

Again, this old video from 2009 explains it best:
https://www.youtube.com/watch?v=3Gmex_4hreQ

Daru925 · Post by **Daru925** » Sat Dec 23, 2017 11:53 pm

Thanks compyfox, i find the -16 setting convenient. I never pushed my music too much anyways.

Googly Smythe · Post by **Googly Smythe** » Sun Dec 24, 2017 1:08 am

I had a quick look at this thread but couldn't find any mention of true peak measurements.
In my utterly amateur fashion, I tend to aim for -11 LUFS (I actually switch Cubase's meter to LUs, so 12 LU - just to be awkward) but I also keep the true peak to a maximum of -1 dB. This gives the conversion from wav to mp3 a fighting chance of sounding pretty good. Because of this, I believe, my Soundcloud tracks sound come across well, mixing/"mastering" errors/misjudgements/lunacy notwithstanding!

poonna · Post by **poonna** » Sun Dec 24, 2017 3:26 am

Compyfox wrote:If you really want to push for CD, a compromise is -14LUFS Integrated (considering that Momentary/MLk can still shoot up to +9LU higher for forte fortissimo passages, which could result -5LUFS MLk max!). But get used to the -16LUFS value, as this is considered an "the best of all worlds" scenario.

According to the K-System, should the MLk shoot up as much as +9LU? I thought it was 3 or 4 LU max (the amber zone for fortissimo passages).

Compyfox · Post by **Compyfox** » Sun Dec 24, 2017 7:20 pm

Googly Smythe wrote:I had a quick look at this thread but couldn't find any mention of true peak measurements.
In my utterly amateur fashion, I tend to aim for -11 LUFS (I actually switch Cubase's meter to LUs, so 12 LU - just to be awkward) but I also keep the true peak to a maximum of -1 dB. This gives the conversion from wav to mp3 a fighting chance of sounding pretty good. Because of this, I believe, my Soundcloud tracks sound come across well, mixing/"mastering" errors/misjudgements/lunacy notwithstanding!

Common ground so far is that True Peak remains at -1dBTP as absolute maximum. This is also a safety net for lossy CODECs like MP3, AAC/MP4 and AC3.

-11LUFS ILk (Integrated Loudness, k-weighted) might be a bit hot for that matter. But the more you push the loudness with a lossy CODEC, the higher the headroom has to be to be save from clipping.

A great tool to check that, is actually Nugen Audio's MasterCheck (Pro Version especially). The Fraunhofer ProCodec (both pro and consumer edition) also feature showing you if your signal is save from clipping per selected CODEC.

WARNING:
Not every limiter is doing ISP limiting in accurate fashion! Always check with a third party meter after the limiter.

poonna wrote:According to the K-System, should the MLk shoot up as much as +9LU? I thought it was 3 or 4 LU max (the amber zone for fortissimo passages).

If you talk about the K-System v1 (aka: Dorrough 40A specs, unweighted RMS meter), then there is a headroom of 4dB. If you talk about the K-System v2 (which is my concept from 2012, based upon ITU-R BS.1770-x), then it's +3LU to +4LU SLk (Short Term Loudness, 3s, k-weighted) max for forte fortissimo passages that I recommend.

If you talk the plain ITU-R BS.1770-x specs (or "presets" like EBU R-128 and ATSC A/85), then the value up to +9LU is meant for MLk values (Momentary Loudness, 400ms, k-weighted). According to the white papers, this headroom is for short(!) noise bursts (e.g. explosion effects).

DJ Warmonger · Post by **DJ Warmonger** » Wed Dec 27, 2017 7:07 am

I think you misunderstood. Club PA systems can deliver 120 dB SPL no problem, but regulations state it should be below 100 dB SPL or so.

Right, but it doesn't change much. For years loudness limits were defined by max SPL and not RMS value, so overcompression was used as a mean to make everything louder - in particular TV advertisements. The LUFS concept introdued to stop that.

Compyfox · Post by **Compyfox** » Wed Dec 27, 2017 11:43 pm

You confuse apples with oranges

Super-simplified:
SPL is measured "post speakers", RMS is measured "in the system"

Also see:
http://www.sengpielaudio.com/calculator-db-volt.htm

DJ Warmonger · Post by **DJ Warmonger** » Thu Dec 28, 2017 1:23 pm

"RMS" stands for "root mean square" and is a purely mathematical concept. From the very page you linked:

All field quantities, like voltage or sound pressure are
always true RMS values, if not otherwise stated.

Compyfox · Post by **Compyfox** » Thu Dec 28, 2017 5:10 pm

So?

"RMS measurement" is still considered "in the box measurement" (especially offline - and don't let me get started on electronics), while SPL is measured "post speakers" (Sound Pressure Level, usually after 1m of distance through the air).

Apples and Oranges still

LUFS - confused.