The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Strategy for achieving 14LUFS release Metering & Analysis Plugins
Old 9th February 2018
  #1
Lives for gear
 
Ain't Nobody's Avatar
 

Strategy for achieving 14LUFS release

Correct me if I'm wrong, but my current understanding is that Spotify and others who use -14LUFS level matching do some version of listening to the loudest section of the song (30 seconds?), and using that (integrated?) reading.

So, my first issue is that I don't know for sure exactly how their measurement will be performed, so their figure and mine could be off a bit.

Second issue is that I've been reading up on opinions that backing off from 0dbfs peaks still matters due to ISPS any time file formats are converted, etc.

Third issue (again, correct me if I'm wrong), but it's my understanding that they will turn it UP if it's BELOW 14LUFS. Unless I'm missing something, this would be a problem if that would cause peaks to exceed 0dbfs. Not clear whether they would somehow then clip it, or just turn it up until peaks hit 0dbfs and stop causing my track to play lower than everyone else's, but either outcome is sub-optimal.

SO....

I'm thinking the best solution is to measure my loudest section, dial it in to -13LUFS, then back it off by 0.5db. Ultimately, that will give a LUFS reading of -13.5. That should give a bit of wiggle room in case their measurement is different, allow them to turn it up a bit if necessary without peak issues, and also allow for any file conversions or other things that might happen to the music downstream without ISP's causing issues.


Do I have this all correct?
Old 10th February 2018
  #2
Lives for gear
 
Ain't Nobody's Avatar
 

Anybody got any insight on any part of the question?

For instance, if you submit something that peaks at 0dbfs, but they measure at -16LUFS, what happens? Do they clip it to get it leveled, or do they just not turn it up at all due to the peaks? Something else?

Also, is it possible to use the exact same measurement? Or a better approximation? Is 30 second sample of the loudest section a good strategy, or something more specific?
Old 10th February 2018
  #3
Quote:
Originally Posted by Ain't Nobody View Post
Correct me if I'm wrong, but my current understanding is that Spotify and others who use -14LUFS level matching do some version of listening to the loudest section of the song (30 seconds?), and using that (integrated?) reading.
I'm pretty sure that the song as a whole is measured, and then the service (Spotify) performs loudness normalization based on that LUFS measurement.
Old 11th February 2018
  #4
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by Henrik Hjortnaes View Post
I'm pretty sure that the song as a whole is measured, and then the service (Spotify) performs loudness normalization based on that LUFS measurement.
Hmmm... I've been told otherwise a couple of times now, and it's a pretty big difference in my case nearly always between overall average LUFS and loudest section.

Anyone have specific info?
Old 11th February 2018
  #5
Lives for gear
 
RRCHON's Avatar
Integrated loudness for normalization is an average of the loudness of the entire track. The algo isn't a straight average as there are gates on signal threshold for measurement and the ballistics (not sure that is the proper term) of rise and fall aren't exactly linear.

It's best to read about yourself though, there are numerous white papers published on the subject.
Old 11th February 2018
  #6
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by RRCHON View Post
Integrated loudness for normalization is an average of the loudness of the entire track. The algo isn't a straight average as there are gates on signal threshold for measurement and the ballistics (not sure that is the proper term) of rise and fall aren't exactly linear.

It's best to read about yourself though, there are numerous white papers published on the subject.
I've probably spent at least 12 hours now reading up on it, and there's a lot of conflicting information.

What appears to be most accurate information I can find as judged by research that apparently went into it would indicate that Spotify in particular is essentially using Replay Gain, and adding 3db.

Only instructions I've located on how to dial in the process involve an entirely offline process in Audacity. That's a bit of a workflow killer, so really hoping there's something that can be done internally in Logic since I essentially master the music and vocals separately within a single Logic project, then use a number of techniques to combine them, so using an offline process to determine an adjustment just means I need to roll back a few steps and reintegrate.

Although I'm also not exactly clear on how it's measured for several other services, it is worth noting that YT appears to be including more vids now in it's -13LUFS leveling.

The reason this may be important, and the question I've yet to find an answer to anywhere is what happens on the various services if your track is too dynamic to level match without peaks exceeding 0dbfs.

For Spotify, YT, etc, the question is, do they clip your peaks and achieve level no matter what, or do they hit the peaks and stop so your track is effectively always quieter than others?

Also not clear as to how dynamics affect ISP issues. Does a more dynamic (less limited) track imply a greater tendency for ISP's if I have peaks too close to 0dbfs? Does the limiting (or lack thereof) on a track affect the need to pull back a bit to avoid overs during format conversions, etc?
Old 11th February 2018
  #7
Lives for gear
 
RRCHON's Avatar
It can be vexing sometimes and not all companies might do the same thing, but some of them (spotify, youtube and amazon music) will put your more dynamic track through a limiter and it doesn't always sound that great.

From my experience:
If you have a lossless version of a song that you've bought and compare it to the Amazon streamed version (nothing against them, it's just what I use) you will sometimes notice some softer sections and delicate moments brought forward to the detriment of track.

I thought YT was at -14LU kind of sad to hear they are -13LU now and so soon. I always through they would go to more DR rather than less, $$ C.R.E.A.M.

A live playback loudness meter will get you close enough, if you want to check loudness offline there are easy to use scanners that can monitor folders for you. I know ORBAN even gives one away for free ( METER — Orban )
Old 11th February 2018
  #8
Lives for gear
 
polybonk's Avatar
Its just an offset. Don't worry about it.

Master like normal. Ie check for technical issues and make each track work together on the album whilst sounding nicely finished.

Only spotify use a limiter to bring level up if you are too low. Other than that nothing is changed.
Old 11th February 2018
  #9
Lives for gear
 
Ben F's Avatar
 

Verified Member
The strategy is not to worry about it.
Old 11th February 2018
  #10
So it appears that Sydney, Australia is somehow excempt from all the panic about how to best adhere to streaming service mechanisms?
Old 11th February 2018
  #11
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by Henrik Hjortnaes View Post
So it appears that Sydney, Australia is somehow excempt from all the panic about how to best adhere to streaming service mechanisms?
Haha. Nice catch. But see... everything works backward there, right? (says the typical American)

Seriously, though, the whole details don't matter thing is generally terrible advice... just like all the folks who said for years to ignore loudness. Might make you feel better, but it was terrible advice for anyone competing in the pop music business. It's just not rooted in reality.

Level matching is a good thing, but the answers to these questions DO matter. If I ignore all of it, and do no limiting, and end up with a -18LUFS track with a highest peak at 0dbfs, it matters A LOT whether or not that gets turned up to match all the other tracks, or if mine is permanently down by 4db every time someone is auditioning tracks because the peak stops it from being turned up any further.

It also matters whether they then just lop off the top 4db of peaks. If they do, I'd be a fool not to do it myself more gently where I control the outcome. Ignoring that makes no more sense than ignoring the physical limitations an peculiarities of vinyl and just pressing whatever type of master sounded good on your system.

Last edited by Ain't Nobody; 11th February 2018 at 06:33 PM..
Old 11th February 2018
  #12
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by RRCHON View Post
It can be vexing sometimes and not all companies might do the same thing, but some of them (spotify, youtube and amazon music) will put your more dynamic track through a limiter and it doesn't always sound that great.

From my experience:
If you have a lossless version of a song that you've bought and compare it to the Amazon streamed version (nothing against them, it's just what I use) you will sometimes notice some softer sections and delicate moments brought forward to the detriment of track.
Exactly the kind of info I'm looking for, thx. Anyone know if any of them just don't turn it up if the peaks don't allow?

I'm thinking overall that -13LUFS might be the way to go if only releasing a single version... which should also accommodate any fluctuations in measurement techniques for -14 outlets. Overall, it would seem I'm better off erring slightly on the side of more limiting than the other way around, though not set on that.

Planning then on dropping down 0.5db (Which would then yield -13.5 pre-leveled). Is this still a good practice and amount for peak headroom?

Also, regarding the offline scanners, do I need to use Replay Gain check for Spotify, but it doesn't apply to others, or...
Old 11th February 2018
  #13
Quote:
Originally Posted by Ain't Nobody View Post
Planning then on dropping down 0.5db (Which would then yield -13.5 pre-leveled). Is this still a good practice and amount for peak headroom?
I have the ceiling closer to -1 dB to counter the lossy formats when the destination is streaming. I'm no expert, but it works for me this way after checking (measuring) a song locally, doing various conversions to lossy formats and bit rates.

Again, I'm no expert.

On another note, could this be of any use to you? Can't vouch for the guy, but... Music Streaming & Loudness Normalization — Ryan Schwabe
Old 11th February 2018
  #14
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by Henrik Hjortnaes View Post
On another note, could this be of any use to you? Can't vouch for the guy, but... Music Streaming & Loudness Normalization — Ryan Schwabe
Missed that one. Reading now...

EDIT: Unfortunately, he ultimately makes the same point everyone does about not squashing too much, but doesn't explore the other side much. He gets into it a bit in the comments, though, and at least on YT seems to indicate that releasing BELOW target LUFS would be a bad thing.
Old 11th February 2018
  #15
Lives for gear
 
SmoothTone's Avatar
 

Quote:
The strategy is to not worry about it.
Quote:
Originally Posted by Ain't Nobody View Post
Seriously, though, the whole details don't matter thing is generally terrible advice... just like all the folks who said for years to ignore loudness. Might make you feel better, but it was terrible advice for anyone competing in the pop music business. It's just not rooted in reality.
The point is that loudness is now the last thing that will grab people's attention. We no longer have to compromise dynamics to compete. Let the engineering practices support the musicality and a good song will sell itself. Now, more than ever, we can focus on just making it sound good.
Old 11th February 2018
  #16
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by SmoothTone View Post
The point is that loudness is now the last thing that will grab people's attention. We no longer have to compromise dynamics to compete. Let the engineering practices support the musicality and a good song will sell itself. Now, more than ever, we can focus on just making it sound good.
You're missing the point. I'm not crushing anything. Quite the opposite, I'm looking to ensure that the music doesn't get clipped or damaged by limiters I don't control downstream... which is exactly what would happen if I took that advice.
Old 11th February 2018
  #17
Lives for gear
 
b0se's Avatar
Ignoring how a streaming service may potentially bork your song doesn't sound like good advice.

I'd rather release something optimised/limited to match the platform requirements. If mastered well (and not butchered for loudness), having a service lower the volume to match higher DR music won't really matter (especially for mainstream stuff). Having a high DR track limited and pushed by an unknown variable (the platform) to match is another ballgame entirely. Potential for non-optimal compression/limiting.
Old 11th February 2018
  #18
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by b0se View Post
Ignoring how a streaming service may potentially bork your song doesn't sound like good advice.
It's bewildering to me. ME's will go on and on about how they dial in minutia for vinyl due to what will inevitably happen to their master once it leaves their hands...

... yet with no sense of irony, suggest completely ignoring the absolute inevitability of what WILL happen to their DIGITAL master once it leaves their hands... even when it is possible in some cases to know exactly what those factors are and how to circumvent them.

As if it's somehow more musical or noble to just let some stranger chop off your peaks however they see fit for the only version 99.9% of your fans will ever hear.

Always know the rules before playing a game with consequences. Ignoring the rules is never good advice.

The Dude does not abide.
Old 11th February 2018
  #19
Lives for gear
 
SmoothTone's Avatar
 

Quote:
Originally Posted by Ain't Nobody View Post
You're missing the point. I'm not crushing anything. Quite the opposite, I'm looking to ensure that the music doesn't get clipped or damaged by limiters I don't control downstream... which is exactly what would happen if I took that advice.
You're right, I did miss that. So, similar to what you're thinking, bringing it up to around -13LUFS integrated as gracefully as you can with a -1dBTP ceiling should guarantee that your track remains free from downstream limiting or encoding nasties.
Old 12th February 2018
  #20
Lives for gear
 
Ben F's Avatar
 

Verified Member
Quote:
Originally Posted by Ain't Nobody View Post
Haha. Nice catch. But see... everything works backward there, right? (says the typical American)

Seriously, though, the whole details don't matter thing is generally terrible advice... just like all the folks who said for years to ignore loudness. Might make you feel better, but it was terrible advice for anyone competing in the pop music business. It's just not rooted in reality.
I'm well aware of how to create a master that is optimised for streaming thanks. In fact, I was part of the team that built the DAB network here in Australia, the backward place.

What I'm saying is that mastering to a specific LUFS number is a moving target. I see poeple asking about this on the mastering forums and facebook page- what about Spotify? i Tunes? You tube? Soundcloud etc...

If I sent a -14LUFS master back to a record label they would laugh and never come back. That's a reality check for you. I make a master that is competitive, sounds better level matched, yet still does not have true peak overs and so it is suitable for streaming / MFiT. If I worried about targets (except -24 LKFS for broadcast) then I am never going to satisfy every service.

So my advise was to not worry about the number...master until it sounds great, use LUFS as a guide (not the ultimate truth), and have the master prepared for most streaming services as to not clip the encoders. Clients are happy, and the master will translate.
Old 12th February 2018
  #21
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by Ben F View Post
I'm well aware of how to create a master that is optimised for streaming thanks. In fact, I was part of the team that built the DAB network here in Australia, the backward place.

What I'm saying is that mastering to a specific LUFS number is a moving target. I see poeple asking about this on the mastering forums and facebook page- what about Spotify? i Tunes? You tube? Soundcloud etc...

If I sent a -14LUFS master back to a record label they would laugh and never come back. That's a reality check for you. I make a master that is competitive, sounds better level matched, yet still does not have true peak overs and so it is suitable for streaming / MFiT. If I worried about targets (except -24 LKFS for broadcast) then I am never going to satisfy every service.

So my advise was to not worry about the number...master until it sounds great, use LUFS as a guide (not the ultimate truth), and have the master prepared for most streaming services as to not clip the encoders. Clients are happy, and the master will translate.
I understand I tend to blather, so I never really expect anyone to read through all of it, but just so we're clear, I don't need the reality check as I have no one to answer to but me and my fans. I can release a master at any level I choose, and have no consideration other than doing what I feel is right for the music overall... which includes acknowledging how it is actually consumed after everyone downstream has their way with it.

All insights are appreciated, and though I claim no inside knowledge of anything, I would tend to agree on the face of it that it appears to be a moving target, and whatever is optimized today may not be tomorrow.

I'm more than willing to take a step back and reconsider the target. Given the information I have available at the moment, it seems unwise to release something with a LUFS lower than -14... probably -13 if the same master will be on YT. In either case, I'm referring to something that is actually half a db or so louder, then turned down to avoid ISP's.

If there are other factors to consider, I'm all ears, but unless I were convinced of a high likelihood of most major outlets moving to something BELOW -14LUFS, that seems like the prudent path.

Truth is, though, it sounds like you're releasing LOUDER masters than I am, so... just out of curiosity, what's an average LUFS that you would release to a major label for a dance-pop artist these days, and if they insist on something above -14LUFS, what is their specific justification?
Old 12th February 2018
  #22
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by SmoothTone View Post
You're right, I did miss that. So, similar to what you're thinking, bringing it up to around -13LUFS integrated as gracefully as you can with a -1dBTP ceiling should guarantee that your track remains free from downstream limiting or encoding nasties.
I've heard justifications for all sorts of figures to avoid ISP's from -0.1db to -3db.

I've tended to settle on -0.5, though I can't claim it's entirely scientific. Is there a specific information source, research, or reason for -1.0?
Old 12th February 2018
  #23
Lives for gear
 
Ben F's Avatar
 

Verified Member
Quote:
Originally Posted by Ain't Nobody View Post

Truth is, though, it sounds like you're releasing LOUDER masters than I am, so... just out of curiosity, what's an average LUFS that you would release to a major label for a dance-pop artist these days, and if they insist on something above -14LUFS, what is their specific justification?
Most major record labels, and many top clients, do not understand or care about LUFS. Never had a request for a target level. You send back a master, the first thing they will do is preview it against other similar releases, probably on a Macbook pro or similar. It has to sound competitive or they will go else where. I'm definitely not the loudest guy out there, my target has always been quality first.

So grab a popular release of a similar genre and test for yourself. for dance-pop these days, it will be probably be -10LUFS to -7LUFS. A decent metering system will show True Peak Overs. If the audio is not heavily clipped, then you may not have to bring the level down by -1dBFS. Heavily clipped material may need more than -1dBFS.
Old 12th February 2018
  #24
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by Ben F View Post
Most major record labels, and many top clients, do not understand or care about LUFS. Never had a request for a target level. You send back a master, the first thing they will do is preview it against other similar releases, probably on a Macbook pro or similar. It has to sound competitive or they will go else where. I'm definitely not the loudest guy out there, my target has always been quality first.

So grab a popular release of a similar genre and test for yourself. for dance-pop these days, it will be probably be -10LUFS to -7LUFS. A decent metering system will show True Peak Overs. If the audio is not heavily clipped, then you may not have to bring the level down by -1dBFS. Heavily clipped material may need more than -1dBFS.
I'm well aware, and I've tested a few tracks. That much is academic to me as I'm really just concerned ultimately with sounding good and making the most of the playing field in which I find myself.

A few years ago, that meant -9 rms masters. Now that it finally looks like that mess will be sorted, I'm looking to instill some sanity and focus on quality... but for me that also means minding what will happen to my masters after I release them.

Trying to remember the commedian's name I saw years ago that said everything is backward in Australia, so when you flush a toilet, it goes counter-clockwise... and up... like a crap tornado. It's probably not half as funny in print, but I got a kick out of it at the time.
Old 12th February 2018
  #25
Lives for gear
 
SmoothTone's Avatar
 

Quote:
Originally Posted by Ain't Nobody View Post
I've heard justifications for all sorts of figures to avoid ISP's from -0.1db to -3db.

I've tended to settle on -0.5, though I can't claim it's entirely scientific. Is there a specific information source, research, or reason for -1.0?
The -1dBTP recommendation is based on the EBU and ITU technical papers that introduced the LUFS metering system. You can search EBU tech for the latest revision of these documents.

Here is the AES paper with recommendations for loudness normalisation of streaming services which draws from the above standards and recommends -1 for peaks.

http://www.aes.org/technical/documen...04_1_15_10.pdf

Also, if you haven't already, you might want to check out Nugen Mastercheck Pro as it's designed to help you work to specific normalisation standards and allows you to audition these along with the relevant codecs.

Just don't forget to enjoy the music as well!
Old 13th February 2018
  #26
Lives for gear
 
polybonk's Avatar
Quote:
Originally Posted by Ain't Nobody View Post
Seriously, though, the whole details don't matter thing is generally terrible advice... just like all the folks who said for years to ignore loudness. Might make you feel better, but it was terrible advice for anyone competing in the pop music business. It's just not rooted in reality.
Quote:
Originally Posted by b0se View Post
Ignoring how a streaming service may potentially bork your song doesn't sound like good advice.

Again its just an offset!

Nothing happens to the audio at all other than level, aside from Spotify only if the level is below -14.
Then it uses a limiter.


If nothing happens to the audio then you can just concentrate on making a great track rather than optimizing for normalization.


Loudness-mal for streaming og download - Lydmaskinen
Old 13th February 2018
  #27
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by polybonk View Post
Again its just an offset!
That just isn't true. EVERY outlet that receives something that is more dynamic than their target LUFS level (below target when normalized) will either:

1) Limit/clip your song as necessary to allow it to be turned up to their target (eg. Spotify)

or

2) Leave yours lower than everyone else's (eg. YouTube)

The only question at that point is WHICH of those two any given outlet will choose since those are the only two options that they have when your track can't be turned up without clipping.

At least be clear when you give someone advice to ignore the issue and submit something not in range that you are, in fact, guaranteeing that their music will be clipped, limited, or stuck lower than all the other tracks on EVERY service.

I find it hard to believe anyone would willingly allow that to happen to their clients.
Old 13th February 2018
  #28
Lives for gear
 
polybonk's Avatar
What are you talking about?

If you submit a track to any of the services listed in that link then the track will be analyzed for LUFS and then normalized to the target level.

This is done with no change to the audio other than gain across the whole track. No limiter applied.

The one exception is spotify if and only if your track is below -14LUFS.

So master tracks as per normal practice and you will be fine.

There is no issue to worry about.
Old 13th February 2018
  #29
Lives for gear
 
b0se's Avatar
Cheers for feedback fellas.

So if you master (well) at say, 12 LUFS as an example, and everyone is happy with the balance (not compromised etc), uploading that will be 'safe', as the only thing that will happen is a volume drop?

I ask as when I use the SilverBullet it's so easy to increase RMS (musically) that I often push past 14 LUFS (13 or 12) and prefer the sound.
Old 13th February 2018
  #30
Lives for gear
 
Ain't Nobody's Avatar
 

Quote:
Originally Posted by polybonk View Post
What are you talking about?

If you submit a track to any of the services listed in that link then the track will be analyzed for LUFS and then normalized to the target level.

This is done with no change to the audio other than gain across the whole track. No limiter applied.

The one exception is spotify if and only if your track is below -14LUFS.

So master tracks as per normal practice and you will be fine.

There is no issue to worry about.
You're either choosing not to pay attention, or just misunderstanding the problem.

Last track I did with no limiting came to a normalized integrated (entire track) measurement of -20LUFS. The normalized bit means -20 LUFS is the loudness reading AFTER offsetting the entire track upwards to hit 0dbfs peaks. "Just offsetting" at this point is not a real world option.

Educate me. What magical process are Tidal, Apple, Amazon, Google, Deezer, and others going to use to turn UP a track 6 more db that's already peaking at 0dbfs?

If you're the ME at that point, it is now your fault that your client's music is clipped, limited, or down 4-6db from every reference on EVERY outlet... actually, on YT your track will now be 7db quieter than everyone else's. That's a hell of a thing to just choose to ignore. The client SHOULD be pissed at that point.

Last edited by Ain't Nobody; 13th February 2018 at 05:55 PM..
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump