Blackmagic Forum

Hello trusted advisors!

I am looking through some tutorials on how to properly set audio levels, I feel that I'm learning more and more, but a couple of things still confuse:

Some people recommend mixing your audio to between -10 and -15db. If you're doing voiceovers, the music underneath, should live at between -20 to -25db as a guide.

Others seem to favor the 0db mark, and mix their audio to be as loud as possible without capping.

I found one article that said for broadcast, -10db is the normal, whilst on youtube it's 0db. THEN I start googling for Audio Levels and YouTube, and I start coming across the term "LUFS" which is new to me, but YouTube specifies -13 LUFS as some form of limit.

I believe the latter is related to Loudness, and so my attention immediately is drawn to the "Loudness" section in Fairlight, however, which of the numbers should I be looking at, and is this still related to my normalization and compression on audio, or is this a different thing to handle?
Here is an image of my current project. I've mixed the audio down to -10db for speech, and -20db for background music.


Any word of a advice to a corona-refugee with way too much time to explore Fairlight?

Some guidance: viewtopic.php?f=21&t=76411#p422973

I haven't gone through it yet, but the Introduction To Fairlight guide below seems like it would be a good resource.

https://www.blackmagicdesign.com/produc ... e/training

Thanks for your recommendations!
As you probably could guess, I'm not a professional, and I'm not getting paid for any of my work. It is merely a hobbyist thing.

The one video that I came across on youtube, which suggested I mix my audio to between -10 and -15db was perhaps thinking about broadcast/TV video. I see that youtube has a -3db limit, and these -13 LUFS. Spotify operates with -14 LUFS.

I will probably read all the material linked by you guys, but it would be nice to just know, for example, if I make a YouTube video, should I then
- Mix it for -3db peak?
- Set the project fairlight settings to -13 LUFS (Youtube specification)
- Now what do I do? Where on the image I posted should I be looking to know that I'm not somehow violating YouTube's loudness specification?

Speculation: If I set my target LUFS to -13, does that mean that the meters will show me variance FROM that? Is sticking closer to zero a good thing? because right now, none of the meters are even close to anything that looks like -13

AND what do I affect loudness with? Is this output level, or some EQ/Frequency thing or what?

Keeping it simple, I set my Loudness metering to A/85 (-24 LUFS) and get the Integrated as close as I can to 0.

For Blu-ray or Digital Copy, I export as-is. For YouTube uploads, I set to the Master to +8 dB which brings the LUFS to the often recommended -16. (Never seen anyone recommend -13, only -16.)

Jim Simon wrote:Keeping it simple, I set my Loudness metering to A/85 (-24 LUFS) and get the Integrated as close as I can to 0.

For Blu-ray or Digital Copy, I export as-is. For YouTube uploads, I set to the Master to +8 dB which brings the LUFS to the often recommended -16. (Never seen anyone recommend -13, only -16.)

My brain needs fresh air. I set my settings like you did, A/85, and made sure project settings Target Loudness Level is set to -24.

I then mixed all my audio to -8db (ish) out of main, with background music being around 10db under speech. Main output now shows between -8db and -14db (ish). Looking at Loudness, I see a value for Integrated of +1.9 for the full clip. How do I know that my LUFS is -16 or close to it? What kind of maths am I missing?

Years ago when I researched this for online content the general consensus seemed to be that the recommendation was to set audio peaks to -6db and leave it at that.

So the first thing I do when adding an audio track is right click > Normalize Audio Levels > Sample Peak Program > -6.0dbfs. If I need to dip the track for an audio voiceover I dip it to -20dbfs but still set the voiceover track to -6dbfs for peaks. When recording talking heads I try to keep the peaks around -6db with a safe track at -12db.

With these settings, I have never had an audio complaint from wedding, music video, and corporate customers. I am far from a sound engineer and audio is my least favorite thing to deal with but these simple settings work for the projects I work on.

herein2020 wrote:Years ago when I researched this for online content the general consensus seemed to be that the recommendation was to set audio peaks to -6db and leave it at that.

Indeed, I'm close to what you're doing, only there are so many different ways of detecting peaks that it is confusing. If you turn on "stats for geeks" on youtube for example, it looks like audio is all over, many "official" artist videos have a normalized audio volume of around -3db. Guess they all try to scream loudly?

The one YouTuber that made sense to me, Jason Yadlovski, made a 44 minute long video named "DaVinci Resolve 16 Audio Crash Course" () that gave me inspiration to look into audio more. He recommends peaks of -10db, which sounds a little soft to me compared to all the other stuff out there.

I'll just look at my main output meter and do my voiceovers/music/sfx to those levels for now, like you do. I might bump it up to -6db too, and try to read up on this LUFS business and what buttons and levers I need to press to adjust them

Pedro Dias wrote:I then mixed all my audio to -8db (ish) out of main

You're looking at the wrong meter. For final mix, use Loudness, not Levels.

Ok, so exploring this further, trying to figure out a practical hands on routine for exporting to YouTube with the correct Loudness:

When I make a video for YouTube (wich says anything over -14 LUFS will be changed / squashed) Here are my steps:

- In project fairlight settings, set the target LUFS to -14
- Mix the audio to -6db (like many here also recommend). This entails the usual stuff, EQ, normalize peaks to -6db, add a little compression on speech. At this point, I'm not paying attention to the Loudness meters, only the Main ouput meter.
- Imported music, which is essentially mixed to peak at 0db, I drag the mixer slider down to a level suitable for mixing with the conversation (around 10db under speech). I usually do audio ducking
- Once the audio is mixed the way I like it, I reset/start loudness metering. and play the full clip to find the loudness of my video (is there any way to speed this up?). Looking at the RIGHT blue bar, I now see a value of 11.5 above it after playing the entire clip, Integrated says +6.9 as you can see below


Here, it stops for me - I have mixed the audio to -6db (on the main meter as you can see from the screenshot). One of the linked posts here explains that LUFS is the right blue bar.

At this point what do I do? Or do I even do anything?
- Will the audio be rendered with a target of -14 LUFS when I export the clip?
- Do I wiggle the Main mixer output until Integrated reaches 0?
- What does +11.4 above the LUFS meter tell me to do?

I have tried to drag down the main mixer level until I get an integrated close to zero, however, it sounds so low that it can't be right, compared to most youtube videos I watch.

I'm at a loss of what I should be doing

I am getting there!

If someone from BLM OR someone else can confirm for me the following, then I think I get most of this:
I found the Loudness graph after reading the manual (yup, takes time, I'm thick headed), and this allows me to visualize the percieved loudness of the clip as such



Can someone verify for me that if I set my LUFS reference level to -14 in the project settings, is it then correct that the 0 on the line represents -14 LUFS?

Victory!

So I compared the exact same video, first without paying attention to LUFS. Looking at it in YouTube with Stats for Nerds turned on I got this:
(Look at the "Volume/Normalized" data)


As you can tell, the volume was "normalized" from 100% to 82%, as it was reported to be 1.7db too loud.

Then I mastered my audio with -14 LUFS as the reference, using the graph in my previous post to check that I'm within normals. I uploaded it to YouTube and looked at the stats:

No normalization whatsoever, the clip uses 100% of the mastered volume without any reduction!

The clips sounded identical to me in level, I suspect that all youtube does is to reduce the master volume on the clip during processing, and since my audio really was just speech with some background music on it, the perception of quality or loss of dynamic range isn't really there, however, if you're mastering content with a high emphasis on good audio, this should mean something to you.

Happy to have this under control now, the aim going forward is to have all my videos at the exact same loudness, without fear of youtube squashing down my dynamic range

This article explains YT normalization.
https://productionadvice.co.uk/stats-for-nerds/

Pedro Dias wrote:Can someone verify for me that if I set my LUFS reference level to -14 in the project settings, is it then correct that the 0 on the line represents -14 LUFS?

Looks like you've figured it out but yes this is correct. Also if I understood your earlier post that you thought hitting the integrated at 0 sounded too low then your monitoring volume is too low. Youtube levels are screaming loud compared to broadcast and streaming services. (unless at that point you were still set to meter target R128?)

Either way glad you sorted it out. The loudness history graph is very helpful.

If you're subscribing to Apple Music, I do recommend a tool called Tunelf Audio Converter. It can help you to adjust the audio level of Apple Music. It could help you remove DRM from Apple Music and convert Apple Music to MP3, then you can freely use Apple Music songs with your program with ease.