MLoudnessAnalyzer

[stewartpeters]

Hi,

I have just bought your MLoudness Analyser.

From reading the EBU Loudness specs, it seems to me that the Integrated level should be around -23dB LUFS. But when I use the MLoudnessAnalyser, the typical Integrated readings are around 9 to 12. I thought the reading would be aminus number at least i.e. -20 or -25, closer to the -23dB LUFS range

Can anyone explain what those figures represent

The Peak and Range figures make sense to me, but I'm still confused with the Integrated meter

Thanks
Stewart Peters
Sydney, Australia

[Ghostwave]

But AFAIK, the EBU R128 integrated meter basic value is 0 (which is a hidden -23 LUFS) so that if you level everything to 0, then it all has the same perceived loudness.
So at +9 to +12, you music is too loud. Turn it down by 9 to 12 dB.

Everyone making music should watch this:

[paterpeter]

The -23LUFS reference is mainly for broadcasting. For mastering/mixing music this is too low for many, so other, higher reference levels are often more useful in practice. For example the "K-System" by Bob Katz which uses -20dBFS, -14dBFS or -12dBFS as reference levels, depending on the program material, or the suggested successor (K-16 v2), which uses a k-weighted -16dBF reference level. See this interesting thread: http://www.kvraudio.com/forum/viewtopic.php?t=369813
Personally, I think K-16 v2 is great and I'll aim for that from now on.

Also, while EBU R-128 says that the integrated loudness should be 0LU+/-1LU, K-16 v2 says that only fortissimo passages should exceed 0LU (momentary loudness k-weighted) and no part should be louder than 3LU. That is, the integrated loudness can differ from 0LU. This concept makes more sense to me for mixing/mastering music.

You can use MLoudnessMeter meter for K-16v2:
- use e.g. EBU +18
- use up to 4LU (momentary) for piano and mezzorforte
- use 4LU - 7 LU (momentary) for forte
- use 7LU - 10 LU (momentary) for fortissimo
- don't exceed 7LU (momentary)
- don't led true peaks exceed -1dBFS

(@Compyfox if you read this - correct me if I'm wrong )

[stewartpeters]

... thanks for the prompt suggestions. But there is something not set right on my meters, because even beautifully mastered albums such as Paul Simon's 'You're the One" and a recent James Taylor album are registering at +10 or +11!

Even a 25 year old Steely Dan album is registering at +4

... but, I notice that changing the 'Offset' to -9 gives much more beleivable readings - around 0 for all these albums - except Steely Dan .

I also checked the levels of all these tracks in another program using a 'K System' meter at K 14 (my favoured setting) and they have believable readings: around 0 to +1

I'm guessing that I'm meant to adjust the 'Offset' to about -9 to get similar readings to what the K System meter is giving me.

Any thoughts?

Thanks

Stewart
Sydney

The -23LUFS reference is mainly for broadcasting. For mastering/mixing music this is too low for many, so other, higher reference levels are often more useful in practice. For example the "K-System" by Bob Katz which uses -20dBFS, -14dBFS or -12dBFS as reference levels, depending on the program material, or the suggested successor (K-16 v2), which uses a k-weighted -16dBF reference level. See this interesting thread: http://www.kvraudio.com/forum/viewtopic.php?t=369813
Personally, I think K-16 v2 is great and I'll aim for that from now on.

Also, while EBU R-128 says that the integrated loudness should be 0LU+/-1LU, K-16 v2 says that only fortissimo passages should exceed 0LU (momentary loudness k-weighted) and no part should be louder than 3LU. That is, the integrated loudness can differ from 0LU. This concept makes more sense to me for mixing/mastering music.

You can use MLoudnessMeter meter for K-16v2:
- use e.g. EBU +18
- use up to 4LU (momentary) for piano and mezzorforte
- use 4LU - 7 LU (momentary) for forte
- use 7LU - 10 LU (momentary) for fortissimo
- don't exceed 7LU (momentary)
- don't led true peaks exceed -1dBFS

(@Compyfox if you read this - correct me if I'm wrong )

[paterpeter]

... thanks for the prompt suggestions. But there is something not set right on my meters, because even beautifully mastered albums such as Paul Simon's 'You're the One" and a recent James Taylor album are registering at +10 or +11!

Those albums are not mastered according to e.g. K-14, let alone EBU R-128.
It's pretty usual for modern tracks to have an integrated loudness of -10dBFS or more. I just measured the first 4 minutes of Deadmau5 - Word Problems (MP3 download version*, released 2009) and it had an integrated loudness of -8.8dBFS, with true peaks of +1.5dBFS. On MLoudnessMeter that would be +14.2LU.

Now imagine how much more room you have for transients/peaks if you're aiming for -16dBFS RMS instead. Hopefully we'll see a shift in the future away from super-hot, peak normalized masters to highly dynamic, punchy masters.



* Legal, payed download, of course! So we're talking about the "official" MP3 version.

[stewartpeters]

... I've just checked a jazz peice on the reference CD from Chesky Records and it reads -2dBFS on my K14 meter, but on the Analyzer it reads +2. Another CD of Bach's Air on a G string reads +5 with the K 14 meter, but my MAnalyzer meter reads +10! on the same track. Do you have any idea why there is there such a big difference?

All the tracks I have tested read up to 5 or 6dB higher on my MAnalyzer than they do on my K 14 meters. Unless I change the offset setting to -6 or -7. Then I get readings that are comparable with K14 readings.

I still get a feeling that I must need to change the offset figure to get something that is comparable to K 14 readings.

The mixes I'm listening to

Stewart
Sydney

... thanks for the prompt suggestions. But there is something not set right on my meters, because even beautifully mastered albums such as Paul Simon's 'You're the One" and a recent James Taylor album are registering at +10 or +11!

Those albums are not mastered according to e.g. K-14, let alone EBU R-128.
It's pretty usual for modern tracks to have an integrated loudness of -10dBFS or more. I just measured the first 4 minutes of Deadmau5 - Word Problems (MP3 download version*, released 2009) and it had an integrated loudness of -8.8dBFS, with true peaks of +1.5dBFS. On MLoudnessMeter that would be +14.2LU.

Now imagine how much more room you have for transients/peaks if you're aiming for -16dBFS RMS instead. Hopefully we'll see a shift in the future away from super-hot, peak normalized masters to highly dynamic, punchy masters.



* Legal, payed download, of course! So we're talking about the "official" MP3 version.

[paterpeter]

In MLoudnessAnalyzer, Momentary, Short-term and Integrated loudness have a reference level of -23LUFS (or dBFS). According to EBU +9/+18/+27 a loudness of 0LU means -23LUFS. When I send in a sine wave of 997Hz* with peak/RMS -23dBFS I get a perfect 0LU for momentary, short-term and integrated loudness and peak/true peaks of -23dBFS.

Input:


MLoudnessAnalyzer:


If I send in a a sine wave at e.g. -14dBFS I naturally get loudness values of +9LU, because (-14) - (-23) = 9

On a K-14 meter it will read 0LU because in K-14 0LU = -14LUFS.


Likewise, a -16dBFS sine gets loudness readings of +7dBFS, because (-16) - (-23) = 7

Again, on a K-16 meter it will read 0LU because in K-16 0LU = -16LUFS.


So if you want to use an EBU meter for the K-System you either have to apply offset or you need to mentally shift the 0LU mark.

If you want to target "0LU":
K-12 => offset -11dB
K-14 => offset -9dB
K-16 => offset -7dB
K-20 => offset -3dB

If you want to keep 0LU=-23LUFS, your targets are:
K-12 => 11LU
K-14 => 9LU
K-16 => 7LU
K-20 => 3LU



If you're looking for a loudness meter that has support for EBU R-128 and K v2, check out ToneBoosters EBULoudness. At 15€ it's very affordable. It also displays the recommended loudness "areas" (dark green, light green, amber, red), including the larger light green area of "K-16v2d(ynamic)" and displays the short-term loudness of up to an hour in a graph, hopefully momentary loudness as well soon.
E.g. feeding in a 997Hz sine at -10dbFS using K16v2:



But MLoudnessAnalyzer with an applied offset should be perfectly suitable, albeit with less visual guidance. The values themselves are accurate as far as I can see. I'm pretty sure Vojtech will update the plugin when he receives feature requests. He's a great developer who listens to his customers.


* That's the proper testing frequency AFAIK.

[stewartpeters]

... yes, that all makes sense now. Thankyou so much for explaining.

I downloaded the demo of Tone Boosters EBU Loudness, but it didn't work properly in Nuendo 3. It loads, but it keeps going into 'bypas' mode. So I'll just stick with MLoudnessAnalyzer.

I can't believe how loud even some of the classical music I tested is!

Thanks again for your time and help

Stewart
Sydney
Australia

[Alphacodex]

I'd love to have the option to see K12 / K14 / K16 meters in then MLoudnessAnalyzer as well as a plain RMS output with 0dB beeing digital maximum at the right boundry.

So here is a

BIG +1

for this feature in a update of the Analyzer. Shouldn't be too hard to implement, though... the basic code for measuring is already in, he only needs to shift the metering and maybe adjust some measurement parameters (i think the intergration time in the K-System is different among others)

Cheers,
Codex

[Alphacodex]

Any plans on adding the option to visually adapt the boundaries in MLoudnessAnalyzer yet?
I'd love to be able to set where the 0 point is and also where the maximum is relative to 0dbFS. I really got used to the K-12v2 / K-14v2 system for mixing and mastering. However with MLoudnessAnalyser the visual representation is always off.

Cheers,
Codex

[MeldaProduction]

Hi,

I have just bought your MLoudness Analyser.

From reading the EBU Loudness specs, it seems to me that the Integrated level should be around -23dB LUFS. But when I use the MLoudnessAnalyser, the typical Integrated readings are around 9 to 12. I thought the reading would be aminus number at least i.e. -20 or -25, closer to the -23dB LUFS range

Can anyone explain what those figures represent

The Peak and Range figures make sense to me, but I'm still confused with the Integrated meter

Thanks
Stewart Peters
Sydney, Australia

I'd recommend using the Target - just set it to whatever you need. For instance spotify and others usually require around -16 LUFS, so if you set the target to this value, then you will want the integrated loudness to be 0 LU. It's pretty low in volume, so many engineers are still reluctant to comply , but these engines will normalize the audio anyways so...