Upcoming major revision of AES TD1004 recommendation on streaming loudness

Upcoming major revision of AES TD1004 recommendation on streaming loudness
0

In line with European initiative by EBU (R128 on normalisation in 2010, revised in 2014 and in 2020), AES is preparing to issue a major revision of recommendation TD1004, initially published in 2015.

This new publication is being redacted jointly with world class audio engineering experts and streaming services (namely Spotify, Apple, Tidal, Qobuz, SoundCloud) who for the most part have already agreed to following the recommendations.

The document is in the last stages of validation and should be available soon. Here is an overview of what its contents should be (subject to last minute changes):

  • New, unified loudness target for streaming: -16LUFS
  • New vernacular: distribution loudness is to be preferred over target loudness
  • Minimum loudness level -20LUFS
  • Short programmes (<1minute): short-term loudness can’t exceed target loudness by more than 5LU
  • Maximum peak level -1dBTP
  • Album normalisation (and playlist normalisation) allowing the preservation of internal, intended loudness differences in an album or playlist
  • New educational website on loudness normalisation to be launched

NARAS (Recording Academy, aka Grammy Awards) will disseminate the recommendation to their members.

One can safely assume that with all the major players being involved in the process, this will put an end to the confusion that has been ongoing about mastering and normalisation best practices lately.

As some of you know, this is a subject that I am particularly interested in and I have been doing some personal research over the past year, which led me to personal exchanges with major experts on the subject, from whom I learnt about this. I thought some of you might be interested.

3 Likes

Very interesting, thanks for sharing your knowledge! Please keep us updated when you find out it is completely finished. Interesting that there’s a minimum loudness level in there. It makes sense, but I don’t think they had that in place previously. (?)

1 Like

Indeed there was never a minimum loudness requirement. My understanding is that the introduction of the minimum loudness level has two goals:

  • improve the audibility of streams on mobile devices
  • reduce the need to turn up loudness too much on softer content (which requires peak limiting, potentially detrimental to the audio)

In my view one of the major changes here is the album/playlist normalisation. One of the few complaints some of us had about normalisation is that you could end up with a loud upbeat track next to a soft track in an album or playlist, and both would sound just as loud, which doesn’t make sense. Actually Spotify went ahead and already introduced this change a couple weeks ago. It is great that the artist/producer intent can be retained with this new feature.

One of the normalisation myths I’ve heard repeatedly is that you should always match the loudness of every single track in an album. So in case anyone is confused about this I encourage you to actually test the loudness of your favorite albums and you’ll see that a lot of them have changes across tracks. Not all albums require the same changes of course. A full-blown EDM or heavy metal album will probably have a high and consistent loudness throughout, but for the majority of genres there is a variety of loudness intensities between the songs in an album, that is what makes the whole listening experience a pleasant one.

Here is an example with this album I recorded and measured, Brothers In Arms by Dire Straits:

2 Likes

Interesting. The Dire Straits example does have quite a bit of variation. I wonder what standard they used when mastering those songs (very pre-LUFS)? Or just used their ears to judge for what sounded consistent through the album.

This album was released in 1985 so yes, that is well before the digital age and LUFS.

At that time, additionally to their ears, they used the VU meters on tape recorders and mixing consoles. VU meters have a slow response (attack and release around 300 ms) and ignore the peaks, so that was one way to have an idea of the short-term average loudness.

Fun fact: a mixer friend of mine told me that the reason why so many records of this era (80s) are so bright is because cocaine affect your hearing in the top end, and pretty much everybody was using it at that time.

2 Likes

Thanks for this great summary Jean-Marc!!

1 Like

Yes, I believe it’s a fairly decent gauge of the RMS value, and that would have likely been their main way to try to hit their ‘target’.

Ain’t that the truth! :blush: I don’t think that fact was included on the “better living through chemistry” flyer. :joy: Multiple other drugs can affect your hearing also, including aspirin/NSAID’s and some antibiotics and antidepressants. Also, caffeine, alcohol and nicotine IIRC. And that’s just a starter list. Likely, for many of us our hearing is ‘off’ its optimal more than it’s ‘on’. :ear: Unless you’re a monk on vegan diet.

2 Likes