Setting LUFS levels

Paul-Fegan wrote on 2/14/2020, 6:34 AM

Hi all, I record and edit a lot of VO audio and for the past 20 years, have been using Normalise to keep levels consistent. However, I have a client who has asked for the audio to adhere to the following loudness specifications:

ITU BS.1770 with a loudness of 16.0 LKFS with a variance of +/- 1.1 and a max peak level of -0.3 dBFS

I understand that LKFS is very like LUFS.

Does anyone know how to set these values for audio in Sound Forge and can it then be saved as a preset and used in a batch file? I know Ozone 9 Advanced has a maximiser where you can target a LUFS level, but before I look into that, I'd like to know if it can be done in SF first.

Many thanks.


rraud wrote on 2/14/2020, 10:09 AM

Your stated 16.0 LKFS would be (minus) -16.0 LKFS. And like ATSC and EBU, it is usually plus or minus 2.0 LUFS, SF does not have an auto level setting for loudness. There are some third-party plug-ins and apps that would get in the ballpark though. Getting the integrated loudness level correct is basically trial and error but relatively easy after some practice. Like TV and mixing for film mixing lock in your monitoring levels. After a while you won't even need to look at meters. Then do a final verification using the 'Statistics" process which analyzes the entire file much faster than real time (or a selected range) and lists the integrated, range, momentary, true peak, ect. You can also copy and paste the statistics as well. Yes, LKFS / LUFS and basically the same.

Paul-Fegan wrote on 2/14/2020, 10:42 AM

Thanks, Rick. And yes, my mistake: minus 16. I see that Adobe Audition has something to deal with this so I was hoping Sound Forge might. No worries. You confirmed what I feared, which is that LUFS require a lot of trial and error, so I'd better start doing some tests. My head is melted. Wish me luck! And thanks again for your swift response.

rraud wrote on 2/14/2020, 12:49 PM

Good luck Paul. I don't know about the iZ Ozone LUFS auto level, but the plug-ins and apps I've encountered worked OK but only had settings for EBU and ATSC TV broadcast (-23 or -24 LUFS).

If you mix/master music for streaming, the Loudness Penalty Analyzer may be useful. It does not auto-set levels though, just like the SF integrated statistics at around -14 LUFS, for YT, Spotifiy, Apple, ect.

Paul-Fegan wrote on 2/15/2020, 9:45 AM

Yeah I'm using trial and error to make settings, but when I run them on a batch, the results fluctuate from file to file. Not much, but certainly not spot-on the target of -16. Apparently Adobe Audition has this built-in, but I don't use it and won't be buying it just for this. Would be good for MAGIX to add this, but for now, I'll make do with the workaround and see if the client is happy. The Statistics window in SF is very useful, but I just need something to help set the levels. It's such a can of worms and I never thought I'd be stressing over volume! :-D I've done a quick search for plug-ins that do what Audition is doing, but have found nothing so far. Most of the results are just meters. Anyway, I'll keep chipping away at the problem here. Thanks again, Rick.

James-Lo wrote on 2/15/2020, 1:08 PM

Just curious: if statistics reports a file to have an integrated LUFS of -17dB, is it just a matter of using the volume tool to raise the level 1dB to reach -16dB LUFS? Or is it more involved than that? What if raising the level 1dB produces clipping somewhere (assuming that's possible)?

rraud wrote on 2/15/2020, 3:49 PM

-17 LUFS should be acceptable if it's like broadcast regs, plus or minus two LUFS. You could raise the volume a full scale dB If it does not clip or even part of the file if a section is slightly lower. if it clips, you could use a brick-wall limiter with a conservative threshold, or tame the peaks manually if they are few. In any case, one point difference in integrated LU measurements is not audible. If the file is 00:30 secs or less the ATSC used to recommended copy and paste to make it a minute or so in length to get an accurate integrated reading.

James-Lo wrote on 2/15/2020, 4:01 PM

Aw, I should've picked a better example. I was trying to see if this kind of leveling might be worth trying to script. But in the case that raising the level to some target requires limiting or compression, I doubt anyone would be happy with the choices a script would have to make. Too many variables, many of them subjective or file-specific, right?

rraud wrote on 2/15/2020, 5:01 PM

Too many variables, many of them subjective or file-specific, right?


Paul-Fegan wrote on 2/16/2020, 5:20 AM

Interesting discussion, guys. What I've found works for now is normalising twice: first time using RMS (have used equal loudness contour, but haven't tried without) and then using peak. I kept doing this with different RMS levels until SF's Statistics panel started showing me the right figures. It's very haphazard and the LUFS level fluctuates a little from file to file, but if you are right about tolerances, Rick, then it should be fine.

rraud wrote on 2/16/2020, 10:04 AM

I'm not aware of any (absolute) -16 LUFS broadcast submission spec. Though I use -16 LUFS for web based (YT and Vimeo, ect ) projects and -14 for music.

Paul-Fegan wrote on 2/16/2020, 11:25 AM

They haven't told me what it's for, but the peak suggested it might be for CD. However, that seems like a pretty antiquated delivery medium at this stage. Anyway, this has all been an interesting discussion and I've learned a few things about the measurement. It would have been much easier had this been one long WAV file, but as it was over 100 shorter clips of narration, I needed to find something that worked across all files in a batch and I'm happy I got close to that. Next time, I might try to get the unedited, raw audio to -16 LUFS before I start editing. It will all depend on the dynamics as to whether or not this is a better method.

rraud wrote on 2/16/2020, 3:24 PM

PCM music files for CD's are usually set to peak 0.01 dBFS many contain some clipping not to mention inter-sample clipping. That said I'd just give the client whatever level they want. They probably just read the -16 LU loudness somewhere. If your VOs are going to be mixed afterwards with other material, the LUFS factor is kind of meaningless.

Paul-Fegan wrote on 2/17/2020, 10:28 AM

Yeah not sure where this is going to end up, but my thought was that they're just audio magazines so may be voice only. I thought I say -0.03 peak when setting up for CD in Ozone, but maybe not. I've sent on a ream of audio so hopefully they're happy with the levels I've set. Thanks for all your help and insight, Rick. Cheers.