Word up! I do a lot of music, sound design, and final audio for Amazon, and -24 LKFS is definitely how we go about things. Thanks for noticing!

very helpful article, but please write on the bit rates and sampling rates for the audio music [say, for mp3 and w/o video] and also for video where audio is present in voice and music..

I recently set up a video streaming/recording system for my church; we publish sermons on Vimeo. I struggled to convince folks that getting the level close to -24 LKFS was important, but we are getting there.

Orban offers a free loudness level meter app that I have used to measure online levels. Using this tool, I have surveyed various online content and found that, for example, Amazon is publishing content at a fairly consistent level around -24.

On the other hand, I listen to an FM talk radio station that streams online, and their loudness is so high that the Orban tool goes off the scale.

Awareness is the first step. Spread the word: -24 LKFS.

Thanks for clarifying the issue. I won’t lie to you – A lot of this stuff goes over my head. There are a lot of technical terms and jargon that make sense when I read explanations about them, but then I forget them over time as I do not need to use them every day.

What I think is clear is that the vast majority of YouTubers do not have a good understanding of audio and how to balance levels (myself included).

Kevin

By the way, is your profile photo of you in front of the Tis Abay? It didn’t have nearly that much water when we saw it.

Of course Kevin, realizing you wrote this article a year ago, and you’ve likely picked up new information along the way. I wrote in the hopes of adding to what you had written so that others who come across your thoughtful post would be able to keep digging with more information.

To answer your question, I can’t make a recommendation as to what level we actually ought to limit YouTube content at: YouTube needs to be the ones to create a published spec for their platform. In reality, the level it’s limited at (which is not the same as normalizing) isn’t that important. Let me explain.

Say I created 2 identical videos, both with audio limited (not normalized, because they’re not the same thing) to -1dbfs. I could then mix both of them (same source material, mind you) so that one played back at -7LUFS and the other so that it played back at -13LUFS. I could make this drastic change in loudness only using the volume faders, compression (the audio kind, not the data kind) and limiting (which is not the same thing as normalization).

It’s the LUFS which is important, because it is the measurement (arguably) that YouTube is using to “sound check” (Apple TM ;) ) their material.

In my opinion, yes, -1dbfs (True Peak) is a good place to set a limiter for YouTube content. Other commenters have suggested various other “Peak” numbers, (notably -12 and -9) but those are numbers which are more suited to broadcast, and even broadcast specs vary wildly as to peak level. The spec that doesn’t vary wildly is the Average Loudness of the “Anchor Element” (usually speech) which for broadcast is very often specified at -24LUFS.

The answer to your question about “one size fits all” is a bit nuanced. If you understand that “limiting” is not the same thing as “normalizing” and that average loudness is exponentially more important than peak level, then I would say it IS a one size fits all solution. As long as your content meets the -1dbfs True Peak and -13LUFS spec, with the information we have about YouTube’s audio level manipulation, those numbers should be about right.

Many thanks for your comment Stephen.

My article was written from the viewpoint of someone who didn’t really know anything about audio and was trying to find my feet along the way.

So do you recommend all YouTube content creators to limit audio to -1dbfs? Is it a one size fits all type of rule or does it really depend on the situation, what is being recorded, and how the video is edited?

Thanks again for your contribution. It’s great to hear from someone who is knowledgable on this subject.

Kevin

One, those of us in post sound don’t use the word “normalization” when we talk about creating content. Normalization happens (sometimes) at the end of the signal chain. YouTube does it. We don’t do it. Nor do we normalize anything using peak measurements. Normalizing only takes the single loudest peak in an audio clip and brings it up to your prescribed level, which does nothing to even out the level across clips/regions/sentences/phrases.

Instead, we “mix”. We adjust the loudness of audio syllable by syllable so that it can be heard clearly and equally. This is done by automating the fader and using your ears. No program or algorithm can tell you how loud something Really is, only your ears can do that.

Related, normalizing to peaks is irrelevant because that gives you no indication of the actual “average loudness” of the material. The only way to even get audio into the right ballpark is by measuring by LUFS, loudness units full scale. You mentioned Ian’s article, but seemed to miss that he wasn’t talking about peak values. He was talking about LUFS. YouTube is now “normalizing” to -13 LUFS (not dbfs).

Lastly, you were on the right digging into izotope’s tips and tricks, but you missed the extremely important term, “True Peak”. As a previous commenter mentioned, inter sample peaks can exceed a digital systems dynamic range causing distortion, even if the loudest sample in that material is less than 0dbfs. So, normalizing to -0.1 isn’t going to help you. The material needs to be “limited” to -1dbfs using a true peak limiter like the one found in the ozone suite, or others.

Once you start thinking about limiting to -1dbfs along with targeting your loudness to -13LUFS, you’ll then be in the right ballpark when publishing audio content for YouTube.

There does not seem to be any hard and fast rules for that. If you look at advice from other YouTubers, they all advise different things.

With no vocals present, you can set the background music to the maximum level if you wish. It is only when it is occurring with vocals that it becomes difficult to balance it all.