• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Loudness wars are over - with lossy Dolby Atmos

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
If they mixed it as loud digitally as they would stereo we're talking about instant complete hearing loss, +150 dBSPL.
They have volume controls in mastering suites, they are not producing these crushed CD mixes at max volume.
 

goat76

Major Contributor
Joined
Jul 21, 2021
Messages
1,269
Likes
1,385
I wouldn't celebrate just yet. For most mastering engineers Atmos is still a relatively new delivery format which means they end up sticking to the guidelines. But as your numbers for the Weeknd release show, Dave Kutch is already working on breaking the equal loudness rules. All the Weeknd tracks have the same LUFSi, and so will pass a simple loudness scan, but he's working on compensating for this by changing the way he limits the audio. This is most obvious for Sacrifice, which has a true peak down at -8dB (!) leading to a miserable crest factor. We'd need to actually look at the audio waveforms to work out precisely how he's mangling the audio for the other two though.

iTunes tried to discourage hard-limiting by refusing to certify audio that clipped at 0dB. Mastering engineers just evaded that by limiting at -0.3dB instead, leading to the same crappy flat-topped clipping distortion, only at a very slightly lower level. I have no doubt that if Atmos becomes a reasonably popular delivery format then engineers will work on ways to break its rules in the same manner.
I can’t see how they will get around the loudness limit of -18 LUFS set for Dolby Atmos. They can of course limit the music as much as they want, but never get around the LUFS normalization level that is based on average volume. At least they will get more headroom for sure. :)
 
Last edited:
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
The ‘Atmos’ you get via a USB DAC on Apple music is nothing but a binaural downmix of the whole thing which sounds really bad on stereo speakers.

Looking a bit more into it.

Apple's Atmos is crapper than Amazon's and Tidal's.

I just measured Apple vs Amazon - same high DR but Amazon sounds like stereo version but with much higher DR.

Apple's does sound really bad as you said. The explanation is given here below.

Dua Lipa's Levetatin (which I measured in OP) is a great example of how crap it sounds on Apple Music with USB DAC.

But then Amazon Music app resamples everything to PCM192kHz in 'Exclusive Mode' lol

I'm using headphones by the way - I haven't listened on stereo speakers.


1647002304295.png
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,423
Likes
7,940
Location
Brussels, Belgium
Looking a bit more into it.

Apple's Atmos is crapper than Amazon's and Tidal's.

I just measured Apple vs Amazon - same high DR but Amazon sounds like stereo version but with much higher DR.

Apple's does sound really bad as you said. The explanation is given here below.

Dua Lipa's Levetatin (which I measured in OP) is a great example of how crap it sounds on Apple Music with USB DAC.

But then Amazon Music app resamples everything to PCM192kHz in 'Exclusive Mode' lol

I'm using headphones by the way - I haven't listened on stereo speakers.


View attachment 191705
I don’t personally find the binaural downmixes bad when you listen to them on headphones.
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
I wouldn't celebrate just yet. For most mastering engineers Atmos is still a relatively new delivery format which means they end up sticking to the guidelines. But as your numbers for the Weeknd release show, Dave Kutch is already working on breaking the equal loudness rules. All the Weeknd tracks have the same LUFSi, and so will pass a simple loudness scan, but he's working on compensating for this by changing the way he limits the audio. This is most obvious for Sacrifice, which has a true peak down at -8dB (!) leading to a miserable crest factor. We'd need to actually look at the audio waveforms to work out precisely how he's mangling the audio for the other two though.

iTunes tried to discourage hard-limiting by refusing to certify audio that clipped at 0dB. Mastering engineers just evaded that by limiting at -0.3dB instead, leading to the same crappy flat-topped clipping distortion, only at a very slightly lower level. I have no doubt that if Atmos becomes a reasonably popular delivery format then engineers will work on ways to break its rules in the same manner.

Amazon Atmos:

1647193674062.png


Amazon Lossless (same as Apple Music Lossless) - loudness norm. obviously off here. Turn on, no clipping but same DR.

1647193719327.png
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
Unfortunately the clipping markers obscure the waveform (turn off View->Show Clipping). It would be helpful to see what the Lossless file looks like with loudness norm turned on to match the output from Atmos. I suspect he might be using a leaky compressor that crushes most of the peaks but lets a few stray ones through.
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
Unfortunately the clipping markers obscure the waveform (turn off View->Show Clipping). It would be helpful to see what the Lossless file looks like with loudness norm turned on to match the output from Atmos. I suspect he might be using a leaky compressor that crushes most of the peaks but lets a few stray ones through.
This is with Apple Music's SoundCheck enabled

I think you're right - crushed peaks compared to Atmos version.


1647196931960.png
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
Unfortunately the clipping markers obscure the waveform (turn off View->Show Clipping). It would be helpful to see what the Lossless file looks like with loudness norm turned on to match the output from Atmos. I suspect he might be using a leaky compressor that crushes most of the peaks but lets a few stray ones through.

Amazon Lossless (no loudness normalisation) with clipping markers turned off

1647197117492.png
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
Atmos version (Amazon's) for the win with this album. And I don't even like Amazon itself.

And the Dua Lipa album. And several more I've looked at since.

Lossy but much better DR.
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
Yeah, Kutch is just shaving off all the peaks. The TPL numbers you got before are probably just a result of the program trying to interpolate the insane clipping, but it doesn’t look like he’s doing anything sophisticated.
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
The ‘Atmos’ you get via a USB DAC on Apple music is nothing but a binaural downmix of the whole thing which sounds really bad on stereo speakers.
I just want to point out that this is apparently not the case if you have a multi-channel USB DAC. If you have that, MacOS will actually output Atmos to your multiple speakers, without the use of an external hardware decoder like an AVR.

This, to my knowledge, is unique, though if the other information in this thread is correct it's probably using Apple's Spatial Audio Renderer and not the Dolby one, which would explain how they can do this because normally Dolby says 'hell no' to providing a software renderer. It would be interesting to compare the results in this case to the real Atmos Renderer.

AppleTV will bitstream Atmos to a hardware renderer like an AVR so it would be possible to A/B test. Rather a pain to set up since you'd also need a Mac, a multi-channel DAC, and some way to switch all channels between an AVR and DAC.
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
I just want to point out that this is apparently not the case if you have a multi-channel USB DAC. If you have that, MacOS will actually output Atmos to your multiple speakers, without the use of an external hardware decoder like an AVR.

This, to my knowledge, is unique, though if the other information in this thread is correct it's probably using Apple's Spatial Audio Renderer and not the Dolby one, which would explain how they can do this because normally Dolby says 'hell no' to providing a software renderer. It would be interesting to compare the results in this case to the real Atmos Renderer.

AppleTV will bitstream Atmos to a hardware renderer like an AVR so it would be possible to A/B test. Rather a pain to set up since you'd also need a Mac, a multi-channel DAC, and some way to switch all channels between an AVR and DAC.

It's a shame Amazon Music app and Tidal Desktop app on macOS can't play Atmos yet.

Those 2 don't do any weird and funky Spatial Audio crap that Apple does.

Apple does their Spatial Audio crapola to suit their headphones of course but not great for any 2-channel with non-Apple headphones.
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
Yeah, Kutch is just shaving off all the peaks. The TPL numbers you got before are probably just a result of the program trying to interpolate the insane clipping, but it doesn’t look like he’s doing anything sophisticated.
Just one more. Another mainstream release.

Much better DR with Atmos version on Amazon Music - proper binaural version without Apple's Spatial audio crap which with Apple Music makes it sound weird on non-Apple Spatial gear.


1647226137950.png


Amazon Lossless, normalization off

1647226370965.png


Amazon Loss, normalization ON

1647226411949.png



Amazon Atmos (much better !!)

1647226433275.png



My wishes:

1. Apple updates their Atmos codec and allows proper binaural with non-Apple gear. Per mastering link I shared above. Maybe unlikely because Apple spatial audio gear is selling like hotcakes and will continue to. Their VR headsets coming too. They want to control that entire Spatial Audio experience with their gear.

2. Amazon and Tidal Desktop apps support Atmos playback

3. Unlikely but TrueHD Atmos comes to all those streaming services, instead of Lossy

In everything I look at, loudness wars are over with these Atmos versions.

Much much better dynamic range even with the really popular stuff you wouldn't expect to have good DR.
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
One more datapoint.

Different album. Again this is mainstream music, not targeted at people that really care about DR.

And again the lossy Atmos version has much better dynamic range than the LOSSLESS versions, no matter if normalization is on or off.

1647227252520.png


Amazon Atmos version (best dynamic range):

1647227045118.png


Amazon Loss - with normalization ON:

1647227190206.png



Amazon Lossless with normalization OFF:


1647227146618.png
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
I wasn't prepared for this.

Amazon's lossy Atmos absolutely smokes the 24/96 versions on Apple Music, Qobuz etc

1647353514104.png


1647353526029.png
 

kongwee

Major Contributor
Joined
Jan 22, 2022
Messages
1,024
Likes
276
Because of LUSF, DR isn't that important for streaming for pure loudness than enjoyment.
 
OP
M

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
Because of LUSF, DR isn't that important for streaming for pure loudness than enjoyment.

Not sure what you mean. Have you seen the wave forms above, even with normalization enabled? Comparing with disabled.
 
Top Bottom