Can I 'cheat' on LUFS by adding silent parts?

[Vlaaing Peerd]

First of all, I think LUFS is a good thing. All that maximising to comparable commercial releases has more than often made my tracks sound worse rather than better.

If I however understand the material and interpret a few tests with audio files correctly, it appears to me, you could increase the level of loud parts by adding silent parts and it will still remain under the same dB level of LUFS.

Is this a correct assumption, or am I missing something blatantly obvious?

I am specifically referring to how media sites such as YT, Spotify, Soundcloud, etc are adjusting the levels.

[RichardT]

No, the algorithm will ignore passages of silence and very low signal levels.

[Hugh Robjohns]

Is this a correct assumption, or am I missing something blatantly obvious?

It depends on the loudness normalisation algorithm being used.

For those that use the BS1770 'standard' algorithm it won't work. I'm not up to speed on which services use what at the moment — it has been an evolving scene — but I'm sure others here can confirm.

Anyway, for those using BS1770, there are processes within the algorithm specifically to make sure that quiet periods don't drag the average loudness value down unfairly.

So, if the signal level falls below -70LUFS — ie. silence — it plays no part in the integrated loudness measurement process. The algorithm effectively freezes for the duration of the silence.

This function was added to later versions of the BS1770 algorithm to make sure that silent sections at the start and end of a programme (as well as pauses in the middle) don't affect the loudness value.

In addition, if the audio level falls 10dB below the current integrated loudness value it's also ignored and the algorithm freezes again, so that pauses in dialogue, or quiet sections between action sequences, dont pull the integrated value down. This element was included in the loudness algorithm to make it focus on the 'foreground' sounds in the material.

In practice, with the BS1770 algorithm its easy to make the integrated loudness value rise, and quite hard to make it fall during the mix process.

[Vlaaing Peerd]

Thanks, with silent I actually meant adding more quiet parts of music, not entirely silent. Without having a look I'm not sure they would be -10dB below the average, I'll get back about that.

I'll have a read on the BS1770 standard, that might provide some insight

[RichardT]

Spotify and Tidal both use the BS. 1770 algorithm.

-10 dB below the average would definitely count in the algorithm.

[amanise]

Can I just say - I'm just totally in awe of the whole idea of LUFS cheating! I am so going to try that out on 'the next one'. The 21st Century equivalent of having pots that go up to 11.

I'm amazed there's a BS for it - it must be a nightmare keeping that current at the moment. Must read it. What a great thread!

[Drew Stephenson]

If you want to cheat on your LUFS then just shave off your bottom octave for the half the song.

[Matt Houghton]

There are certainly ways to write/mix music that results in them being perceived louder without measuring louder. For example, having a loud part at the beginning to catch the ear after the previous song's outro, before taking the listener on a journey to wherever the song needs to sit. Or using shorter punchier sounds instead of long sustained ones (more melody/counterpoint, less by way of organs, pads and long reverb washes etc), or using ear-grabbing effects like delays and tremolo. But these all have a significant impact musically. And surely your primary goal is to make good music...

Adding silence/quiet sections won't help much. For one thing, there's the gating function described above, which doesn't just filter out silence. For another, at best, such tactics will only deliver a few of tenths of a decibel of change in the integrated LUFS reading over the course of a typical track, which is pretty much imperceptible. Unless you take the piss and ruin the song, that is...

Better, IMO, just to make music that sounds good and ensure it isn't *too* quiet (such that the true peak prevents algorithms from raising the level to at or around the 'target'). If crushing things with limiters and clippers is a nice sound so be it. If it isn't, don't do it. Simples.

Still... if you have these sort of ideas, why not just try them out for yourself and measure the differences in LUFS and do some blind tests to see what you think of the differences subjectively? Just how long a 'quiet section' would you need to graft on to a typical 3m30s pop track to make an appreciable difference to the LUFS-I, and what are the impacts subjectively in terms of perceived loudness and musical appeal?

[RichardT]

As Matt says, a trick to maximise perceive loudness for a given LUFS value is to avoid sustained sounds. You want to maximise percussive peaks and minimise the level between them.

[tea for two]

There are certainly ways to write/mix music that results in them being perceived louder without measuring louder. For example, having a loud part at the beginning to catch the ear .... before taking the listener on a journey to wherever

Da Da Da Duuuuum
Da Da Da Duuuuum
one of the most famouse examples of this.

Or using shorter punchier sounds instead of long sustained ones

Dum Dum
De Dum Dum
hook from Soft Cell : Tainted Love.

Mint Royale : Sexiest man in Jamaica, the hook is some bloke warbling "I'm the sexiest man in Jamaica"

I think can make the hook stand out whichever way including

using ear-grabbing effects like delays and tremolo
If crushing things with limiters and clippers is a nice sound so be it. If it isn't, don't do it.

[Vlaaing Peerd]

Thanks for all the replies, much appreciated!

I was wondering about this, as I still struggle with especially EDM productions to get it equally loud (in perception) with commercial ones, even though average dB in LUFS are the same and suspected these elements would make it louder.

So effectively:
- adding quiet passages, as long as they aren't less than -10dB below average;
- shortening loud elements in the mix (bass, kicks);

Can increase the perceptive loudness while having the same level of LUFS, correct?

If that is the case, the loudness war isn't really over, it just appears to take place in a new battleground.

[Vlaaing Peerd]

Still... if you have these sort of ideas, why not just try them out for yourself and measure the differences in LUFS and do some blind tests to see what you think of the differences subjectively? Just how long a 'quiet section' would you need to graft on to a typical 3m30s pop track to make an appreciable difference to the LUFS-I, and what are the impacts subjectively in terms of perceived loudness and musical appeal?

Well, until so far the idea just popped up and decided to ask here. I find the task of mastering horrible and generally don't like to touch my mix too much.

But not mastering results in much lower perceived levels compared to commercial equivalents, even if I maxed it to the same Spotify/YouTube/Soundcloud accepted LUFS levels.

Anyhow, I am definitely going to have a try on one track, the original, one with a longer quiet section in the middle, one with shortened loud sounds and one with both longer quiet section and shortened sounds.

If anyone is interested I can post the results here though it might take over the weekend or longer (the weather in NL is really good lately and has an inverse relationship with time spent in my home studio).