Video: Why Dolby Atmos Music at Home Is Always Worse Than In The Studio (No Matter What)

[jimfisheye]

I just did!

The encoded TrueHD+Atmos matches the mix off the mixing desk. (Or matches the mix off the Dolby renderer if you're using that to do the final panning of some of your mix elements as objects.)

Think of the objects as an intermediate step just like the 24 track tape. (Or 500 track DAW session.) No consumer gadget sees that part. It's already baked in together in a mix during production.

So now some mix elements are kept in separate tracks called objects. The Dolby decoder is programmed to speaker manage and do the final mix assembly based on the available system. If that goes wrong... well there it is. Yeah, an opportunity for your mix to come out wrong in a consumer device... Great idea! That aside, so far my encodes sound just like the final mix before encoding.

Show us one that screws up! Post the original 7.1.4 rendered mix. Post the encoded copy.

 

[SeeMoreDigital]

Well I'm convinced. Thanks @jimfisheye

 

[jimfisheye]

Well I'm convinced. Thanks @jimfisheye

This was my very first question. "Is the final encoded file 1:1 with the master render?" Getting a straight answer and even getting clear answers about the software hierarchy was like pulling teeth! I got a lot of defensiveness and misdirection especially about what roles the Dolby renderer app played vs the Dolby media encoder.

Dolby renderer: Produces an intermediary .atmos file set from the master mix render. Can NOT produce the encoded TrueHD+Atmos files. However, the lossy DD+JOC output is allowed.
Dolby media encoder: Can produce the encoded TrueHD+Atmos file from and only from the .atmos intermediate file. Can NOT produce the encoded TrueHD+Atmos file from the mix itself in raw wav audio or any lossless format (eg wavpack).
Dolby reference player: The only software media player in the world at present with the Atmos decoder codec. Plays TrueHD+Atmos mlp files.

To be fair, Atmos' ability to render the final mix (bed + objects) to whatever the connected speaker array supports is a bullet point they want to talk about. In my thinking, and perhaps others around here, I want to know if the main feature of delivering a lossless 7.1.4 mix 1:1 works first and foremost.

The software hierarchy requires all 3 apps. They thought this through! There are a couple 3rd party renderer apps now. I haven't tried them yet. The encoder and the reference player are still proprietary.

So their renderer app makes is easy to monitor in stereo and produce a lossy streaming version of your master. You have to jump through hoops and get approved by the Dolby corporate overlords to be permitted to purchase the encoder license. Make of that what you will.

 

[SeeMoreDigital]

All I will add to this topic is...

The way Atmos audio is encoded for use with Blu-ray audio discs (and streaming) is quite different to how it's encoded for use in cinemas...

My brother-in-law used to work for Odeon Cinemas and was instrumental in the Dolby refit at Leicester Square

 

[jimfisheye]

All I will add to this topic is...

The way Atmos audio is encoded for use with Blu-ray audio discs (and streaming) is quite different to how it's encoded for use in cinemas...

My brother-in-law used to work for Odeon Cinemas and was instrumental in the Dolby refit at Leicester Square

What have you seen with that?

I know there are some audio interfaces aimed at theater use that have the Dolby Atmos decoder built in. (Like the consumer aimed AVRs and other devices.) 16 or more balanced output channels... so all those DACs and then the Dolby codec with the $$$$ price tag you'd expect. So those would live decode the standard encoded Atmos sent from the computer 'pass through' style.

We can just use wavpack (or even uncompressed wav) for 12 or 16 channel files that are just straight audio masters with no encoding and play in nearly any software media player. There are some roadblocks to that for consumers - this was thought through too. AVRs with no ability to input 12 or 16 channels either analog or from a computer connection. MacOS with the approved device whitelists and disabling features and outputs sans approved device connected. Whatever Microsoft is doing.

 

[JediJoker]

The encoded TrueHD+Atmos matches the mix off the mixing desk. (Or matches the mix off the Dolby renderer if you're using that to do the final panning of some of your mix elements as objects.)

Can you do a null test to confirm there is actually zero difference between the two?

 

[jimfisheye]

Can you do a null test to confirm there is actually zero difference between the two?

There it is! I was waiting for that.

Disclaimer: I've been in the mode of wanting to give accolades for making 12 channel mixes a forever thing. Most creative audacious thing I would have never imagined happening in this plastic corporate world! And then saving the salt for the greedy approach with the decoder and the use of that to force hardware sales along with some disdain for the openly lossy streaming variant.

The answer is no. Encoded TrueHD+Atmos does not null with the master. It's actually slightly lossy. Stay with me though! It's more akin to dts2496 level of lossy. (Properly decoded dts2496. This is also a ringer to talk about because the errant core-only decodes sound significantly more damaged than normal lossy degradation. Read the comments from older hardware or software users dinged by that in the Tull reviews.)

A/B test roundup:
Comparing a master encoded to TrueHD+Atmos vs encoded to lossy DD+JOC is blunt. Well outside any perception bias.
Comparing the encoded to TrueHD+Atmos vs the actual master sounds identical. It doesn't null. But every mix or test file or test tone I've rendered sounds the same. Yeah, you'll have to trust my subjective opinion there! Or try it yourself.

They're generating difference signals for the metadata delivery of the object elements. That's how the "objects are delivered in the metadata". The nuts and bolts of that is object elements are mixed into the bed tracks and the difference signals are used to null them out and isolate them after. This is clearly getting into FFT tricks - like doing center extraction from stereo. There isn't a passive phase trick like swapping between mid-side and stereo for that. Their exact method is their trade secret with this.

My usual SOP is to avoid any lossy compression just because 24 bit lossless is a thing and there's no reason to go back or get into vetting mixes and deciding what level of lossy is still transparent. 24 bit wav/flac/wv - just do it. In the spirit of picking my battles... they apparently have this working! At least as well as center extraction trickery. They're not compressing the data, they're doing phase tricks to null elements out of the bed. It's about the ability to go back and forth with the objects rather than reducing data.

But this guy in the video is gaslighting with pretending the TrueHD+Atmos format doesn't even exist! The lossy JOC version is very audibly flawed. OK sure, I've heard worse damage than that by light years! Volume war mastering, for example. Or just a bad mix. Some of the Floyd bootlegs I listen to! So I hesitate to read them the riot act, just like I give those Tull DVDs a pass. I don't hear difference or loss. Putting the A and B renders in DAW tracks and seamlessly A/B'ing between them. (ie. No chirps, clicks, gaps, or pauses clicking between A and B.)

Well, there's the can of worms opened up.

 

[MidiMagic]

Is that lossy or lousy? I'm not really worried about lossy. I am worried about lousy.

 

[PolkSDA]

Welp, there it is, boyz.... tear it all down and go back to mono.

 

[jimfisheye]

Here's the thing. If you really are OCD about this and chasing the "best seats" for a recording, you can vet this stuff yourself. A lot of conversion software is free. Convert something HD that you also think sounds audiophile to different sample rates, formats, etc and listen for yourself. Gather a batch of different releases for a favorite album and compare them after matching volume levels to really hear different masterings.

Gathering the Atmos software is not free and not easy. But you need to do that to even play the stuff in the first place. So run a couple demos of that for yourself while you're at it.

12 channel mixes are real. It's all 24 bit audio. Encoded Atmos delivers it intact. You want this! There's some bs to wade through but I think it's worth it. Attacking the audio and misrepresenting the format like the guy in the video is misguided. Give them shit for the greed with the decoder, yes! But 12 channel mixes are real and you want them!

Yeah, there are some lousy mixes out there. (Did I make a typo back there?) That's what makes the good ones stand out!

 

[jimfisheye]

Welp, there it is, boyz.... tear it all down and go back to mono.

A soundbar will do that for you.

 

[pat bateman]

I will always appreciate inside baseball information, but this is like a news article with a sensationalized headline and then a 1 sentence ringer that makes it not so juicy. He says about midway through the video that a consumer wouldn’t be able to tell the difference if their life depended on it. If that’s the case, why make a video with such strong conviction overtones explaining that it doesn’t sound the same?

 

[zdjh22]

I will always appreciate inside baseball information, but this is like a news article with a sensationalized headline and then a 1 sentence ringer that makes it not so juicy. He says about midway through the video that a consumer wouldn’t be able to tell the difference if their life depended on it. If that’s the case, why make a video with such strong conviction obertones explaining that it doesn’t sound the same?

It was definitely a strike and a miss by Michael G Wagner, who often generates interesting videos that have been worth my time. Not this one.

Perhaps he was just following the maxim “no publicity is bad publicity”, enjoying the revenue from outraged views as much as from other views.

I have never had an expectation that what I hear on my systems matches what someone else hears, in particular, the sound engineer in the mixing or recording studio. It’s almost guaranteed in Atmos that what I hear doesn’t match everyone else, since the renderer/decoder only uses a nominal position for the speakers as defined by the layout (5.1.4, 7.1.4, 9.1.6, etc.) rather than the actual positions. My speakers are not in the canonical locations because my room doesn’t allow it. At best my AVR adjusts for distances to the speakers, and for ceiling height for upward firing speakers, but not for angles.

 

[MidiMagic]

Welp, there it is, boyz.... tear it all down and go back to mono.

Naw. Just back to QS.