Maximising Audio Levels for YouTube – Loudness Without Braking the Rules

the youtube audio level test
We did the test. Loud audio versus Normal audio on YouTube.

Creating video content has never been easier. That is fantastic on the one hand and tricky on the other. Being able to easily create videos and distributing them online, enables virtually everyone to get their message across. The negative is that the amount of content created on a daily basis, makes it more difficult to stand out and get views for your video. Statistics show that on Youtube alone, every minute 300 hours of video material is uploaded and nearly 5.000.000.000 videos are watched every day by roughly 1.300.000.000 users.

Creating appealing visual content is one thing, completing it with a cracking soundtrack is our thing.

Audio is the Achilles heel of many video productions. It’s often an afterthought. In our opinion a missed opportunity to create a truly great production.

Here at SWT Mastering, our customers ask us on a daily basis to master their tracks as loud as possible. Statistically the majority of the productions we master are launched directly on YouTube Channels. Therefore we find ourselves often explaining to our customers that it is useless to go louder than the -13 LUFS that YouTube specifies. Still, many believe that louder is better.

Instead of going through the somewhat boring technical details, we decided to create a small testcase to find out what works best with YouTube.

Firstly we took an instrumental track, composed by SWT’s Levente Kibedi, and finished it in 2 different ways. The 1st being leveled to the YouTube specifications and verified to not distort on any of the codecs used for streaming in different bit-rates. The 2nd one we mastered more to the taste of our clients. (In other words: loud). We then created 2 separate videos, each containing nothing else than the music track, and uploaded them to YouTube.

Video 1

The V1 video is the one according to YouTube specifications.

 

Video 2

The V2 video is the loud(er) one, reaching an average of -8 LUFS.

 

For this test, and because the majority of videos are consumed on mobile devices, we played the videos back on an iPhone 7 plus and measured the volume by using an iPad PRO with a decibel meter app. (Maybe not laboratory level science, but we made sure to measure the videos’ sound levels in the same room with the same volume level setting (75% volume level) and the same distance between the player and the audio meter.) We found out that V1 came to an average sound level of 68,4 dB and a maximum level of 72,1 dB. When we played the second video, the average sound level of V2 came to 69,3 dB and the maximum sound level to 72,4 dB.

For reference we also measured some other videos on YouTube in the exact same setting as described above.

Katy Perry – Chained To The Rhythm

Average dB level: 68,3
Max dB level: 72,6

 

Taylor Swift – Bad Blood

Average dB level: 67,9
Max dB level: 72,0

 

Ferro – Hatza, Hatza

Average dB level: 67,8
Max dB level: 71,6

 

As we where more or less expecting, it turns out that the heavily limited and compressed V2 track does not sound notably louder than V1. We where however somewhat surprised to find out that the reference videos where slightly softer than our test videos. We need to remark here that a track with vocals can be perceived as being a bit louder than our instrumental track because of the frequency range of the human voice and how small speakers will reproduce those frequencies better.

As an experiment we tried to make our track sound a bit louder by applying a few modifications, to see if we could at least influence the listeners perception of loudness. (We’ve already established that making it actually louder doesn’t really work.)

Video 3

We filtered out a portion of the low frequencies to give the mid-range more space and then remastered it to the YouTube specifications. Now with a max peak level of around -1,5 dB. This however distorts on codecs for streams lower than 128 kbps.

Average dB level: 68,5
Max dB level: 71,6

 

Video 4

For this version we altered the cut in the low frequencies and applied some more EQ changes, before mastering it again to the YT specs. As with V3, a max peak level of -1,5 dB.

Average dB level: 68,5
Max dB level: 71,3

 

Although V4 does not register the highest AVG and MAX levels, in our opinion, it sounds more alive then any of the other versions we’ve made. But that is, as it many times is when it comes to audio, our personal perception – taste and ears differ.

In Conclusion

The V2 version is heavily compressed / limited and sounds a lot duller than the V1 version. It indeed registered marginally higher AVG and MAX values but the “price paid” seems to be disproportional to the gains. Besides all the technical talk, this is the point that we want to get across more than anything: take the technical limitations of the distribution platforms you’re going to use in consideration when you’re recording or mixing a song. Why crush it to death when it will not result in a louder track?

Make your work sound open and fresh – more alive. The listeners will love you for it. Also, most digital distribution platforms take an average LUFS value as measurement of volume. Making sections softer will give you the space to go louder in others. Not having to compress the audio as much, gives you plenty of room to play with louder and softer elements in your composition. Famous DJ and music producer Armin van Buuren, explains in one of his Master Class videos how he lowers the volume of the song by a few dB before the drop and then comes in at full volume. Creating a lot more drama. Keep this in mind when creating or mixing your next song. And please – keep it in mind when asking for a louder master… 😉

If you have questions about creating great sounding YouTube video content, get in touch with us. We’re happy to put our high-end analogue mastering studio to work and optimise your music for digital distribution.

1 Comment

Comments are closed.