Feedback on my first Atmos mix (DD+ mp4 download)

[LTW]

To commemorate the 10th anniversary of Look To Windward's "Kepler" EP from 2014 I have done a full stereo remix of the title track AS WELL AS my first Dolby Atmos mix. I would love feedback on the mix from whoever is interested and able to play it back on their system.

This has been mixed on my new 7.1.4 setup (expanded from 5.1). It's a pummeling 11 minute prog-metal epic with a small section of "death growls" which might not be everyone's cup of tea. I also think it's one of my best songs . Dry drums, rhythm gats, bass and lead vox have been mixed into the 7.1.2 bed while a substantial 75 objects constitute the rest of the dense mix. I've had so much fun mixing this. There is a lot going on in this song and it's really come alive with a 3D sound field.

I'm just very curious to find out how it sounds on other people's systems as my reference playback has only been my mixing setup and my 5.1 home theater. I think there are some great setups out there that could give it a test drive. I'd love feedback not just from other Atmos listeners but even 5.1 or other setups. I'm pretty satisfied with how it mixes down to 5.1 and Binaural.

I'm sharing a mp4 file with a Dolby Digital Plus JOC encoded Atmos mix. I'll eventually be releasing it on streaming and as a lossless download from IAA. Note that this is at -18LUFS so playback is quiet.

(Edited link: updated to WIP v3)
https://mega.nz/file/DMF2iSTS#CT06hIygiflTjjjVAindOnBdZIG6hLjtJ7ofHDRszoQ

 

[jimfisheye]

I'll bite!

I like the prog elements. Right in my wheelhouse there!

I'll be blunt. The extreme volume war levels are mutilating! You're at around -10 LUFS and then with that just turned down. (Also using a -19 dialnorm to turn it back down.) Audible distortion from the first note. The drums have no punch or dynamics and everything sounds choked. And way beyond what the lossy treatment imparts.

The mix should sound good upstream from all that and I bet it does. The quiet parts sound right. Things still sound under control in good ways - the mix is still there with a lot of things on point but it's like listening to a brutal mp3 copy of something.

I'd set the monitoring for -14 LUFS (full 12 channel metering.) listening levels and let the mix breath. Nearly all of the commercial Atmos mixes I've heard so far are in the -16 to -13 LUFS range. Often normalized down to -16 to -18 LUFS range with a straight normalize reduction (not compression).

I just listened to the 5.1 track you posted too. This one is way more under control but it also sounds a little choked when the loud section kicks in. Sounds pretty perfect up to that point. The drums especially again. Just back off on levels and let people use their volume control and these mixes would be fully badass!

7.1.4 system here using the Dolby reference player.

 

[LTW]

I'll bite!

I like the prog elements. Right in my wheelhouse there!

I'll be blunt. The extreme volume war levels are mutilating! You're at around -10 LUFS and then with that just turned down. (Also using a -19 dialnorm to turn it back down.) Audible distortion from the first note. The drums have no punch or dynamics and everything sounds choked. And way beyond what the lossy treatment imparts.

The mix should sound good upstream from all that and I bet it does. The quiet parts sound right. Things still sound under control in good ways - the mix is still there with a lot of things on point but it's like listening to a brutal mp3 copy of something.

I'd set the monitoring for -14 LUFS (full 12 channel metering.) listening levels and let the mix breath. Nearly all of the commercial Atmos mixes I've heard so far are in the -16 to -13 LUFS range. Often normalized down to -16 to -18 LUFS range with a straight normalize reduction (not compression).

I just listened to the 5.1 track you posted too. This one is way more under control but it also sounds a little choked when the loud section kicks in. Sounds pretty perfect up to that point. The drums especially again. Just back off on levels and let people use their volume control and these mixes would be fully badass!

7.1.4 system here using the Dolby reference player.

Thanks for taking the time to listen!
I'm immediately concerned with loudness being an issue because I mixed this to -18LUFS Integrated. It sounds nowhere near distorted or limited in my DAW, and I am not using any limiters or master bus compression on the mix.

I checked the DD+ mp4 in reference player myself and I noticed that changing the Dynamic Range Control setting has a massive affect affect on the playback loudness. At the default 'Line' setting it plays at the correct loudness but when I change it to 'Disabled' it jumps right up to -10 on the LR channels. I don't really understand how DRC works with Dolby encoding or playback and it seems to add another level of complication to the process. I'm reasonably confident my mix is at the Dolby recommended level of -18.

As for the distortion, I can definitely hear it introduced by the EAC-3 encoding. It's not there in the original mix or the TrueHD encode. But when played back with DRC 'Disabled' in Reference Player there seems to be even more distortion added. If you have a moment, could you check it using the 'Line' setting for DRC in the Player? Just for my own sanity! Perhaps there's also another setting I should be using when encoding to DD+.

The drums sound quite punchy to me, and despite it being a dense with distorted guitars, still pretty dynamic. I hope the issues you found are due to an encoding/playback issue and less from my mixing! It's very helpful to discover these things though so thanks again for the feedback!

 

[LTW]

@jimfisheye I don't know if I have discovered the cause of the loudness or distortion but I re-encoded the DD+ mp4 with Content set to 'Music' which I had missed last time. Also set the DRC setting to 'Music Light' instead of Standard. No idea if that makes a difference. I can't really hear any distortion now in Reference Player. I updated the link above.
Here's how it looks for me:

And these are the decoded channels I got out of MMH Atmos Helper. Rather quiet! I'm trying to play detective at this point. Figure out if I'm doing something wrong:

 

[jimfisheye]

Oh, so something weird is going on.

I haven't gone over the lossy mode extensively. I focused on the lossless delivery and verified that from a few different directions now. So I don't mean to be unhelpful but I might be.

The v2 sounds a little brighter. Something still sounds more compressed than it should though. The levels don't make sense to me with a dialnorm of -19 either. This is at approx -12 LUFS absolute level. Rock appropriate loud. It's normalized to -17 with a -5 peak. Nothing should sound wrong from any of that. Normal music dialnorm is -31. -19 should be 12db quieter. So how does that add up here? The compressed effect might be in line with 7 or 8db of hard limiting.

No compression settings enabled in the reference player.

Can you share a straight wav or wavpack 12 ch file? (If that's easy.) Or mlp or mlp(thd) in mka.

PS
1st time I took a level sample in the middle and got -17 with -7 peak. Clearly limited peaks and all so an absolute level of -10 LUFS normalized. And the extra compressed sound and apparently more dulling as well all sounded volume war like. So that's what I said the first reply. Looks like the whole thing is more accurately -17 with -5 peak though. The compressed sound shouldn't come from that.

 

[LTW]

Oh, so something weird is going on.

I haven't gone over the lossy mode extensively. I focused on the lossless delivery and verified that from a few different directions now. So I don't mean to be unhelpful but I might be.

The v2 sounds a little brighter. Something still sounds more compressed than it should though. The levels don't make sense to me with a dialnorm of -19 either. This is at approx -12 LUFS absolute level. Rock appropriate loud. It's normalized to -17 with a -5 peak. Nothing should sound wrong from any of that. Normal music dialnorm is -31. -19 should be 12db quieter. So how does that add up here? The compressed effect might be in line with 7 or 8db of hard limiting.

No compression settings enabled in the reference player.

Can you share a straight wav or wavpack 12 ch file? (If that's easy.) Or mlp or mlp(thd) in mka.

PS
1st time I took a level sample in the middle and got -17 with -7 peak. Clearly limited peaks and all so an absolute level of -10 LUFS normalized. And the extra compressed sound and apparently more dulling as well all sounded volume war like. So that's what I said the first reply. Looks like the whole thing is more accurately -17 with -5 peak though. The compressed sound shouldn't come from that.

Perhaps it's just the nature of this kind of music. Distorted guitars are inherently compressed and drums generally have a fair amount of processing to get them to cut through a mix. I also really dislike the sound of overly slammed and distorted mastering and I master my stereo mixes relatively quiet for this genre. But there is no mastering at all on this Atmos track.

Here's Kepler (with +14db gain added so you can see the waveform) next to SW's Atmos mix for 'Harridan'. Same kinds of peaks for snare and kick hits. No hard limiting.

Link to the first 2mins of Kepler as a 12ch decoded wav (from the TrueHD Atmos)
https://mega.nz/file/zBlkwAzb#yjl52mMHS6M1qdMiO2zPI2n9MjwM6jDlnUlIpcHSQHs

I think if I am below the -18LUFS threshold for submitting to streaming services I should be in the clear. IAA take care of the Dolby encoding on their end too so if the ADM is clean I should hopefully be good.

 

[jimfisheye]

Perhaps it's just the nature of this kind of music. ...

Very much no!
I'm very familiar with the styles and sounds in this. I'm hearing hard limiting and lossy artifacts. The sound is choked out and it's severe.

Hearing the actual wav file would let me critique the intended mix and maybe spot further clues leading to the mystery. I understand if you're not willing to share the lossless copy though. Or PM me.

 

[LTW]

Very much no!
I'm very familiar with the styles and sounds in this. I'm hearing hard limiting and lossy artifacts. The sound is choked out and it's severe.

Hearing the actual wav file would let me critique the intended mix and maybe spot further clues leading to the mystery. I understand if you're not willing to share the lossless copy though. Or PM me.

There's a link to the 12ch wav above, decoded from the TrueHD encode. Any channel based wav file will always need to be a decoded from the object based mix. The original ADM wav is 7GB which I won't be sharing

 

[Thonolan]

I'd love to give you my feedback since I love prog metal and growled vocals are absolutely not a problem for me (I'm more of a prog rock guy, but I do enjoy lots of extreme metal too), but my Atmos system isn't ready yet. The room is almost finished and I may be able to hear it in 2 or 3 weeks. I cannot use my old 5.1 system either, as I'm reforming that room as well and only stereo playback is available for the time being.

 

[jimfisheye]

Just grabbed the wav file. Give me a minute to coffee up and venture into the studio.

 

[jimfisheye]

The wav file sample is free of the lossy sound! The mix balances are different and sound more intentional.

Something is still very wrong...

This sample is at -32 LUFS with -17db peaks. So... normalized down by 17db! What the heck? Still an 18 bit recording. Still sounds more intentional. But something is going very weird with your connection with the Dolby renderer. Or something happened with the round trip.

That -5db peak level I saw with the lossy copy... Turning that whole thing down by another 12db might match up with that -19 dialnorm setting (which would turn everything down by 12db. Straight normalization with that. Not compression.) But I don't think this speculation is useful. Other parts of that don't add up.

Are you doing some final panning or something in the Dolby renderer? Is your mix from the DAW incomplete with something like that going on?

I use the Dolby Atmos system as a delivery accessory. I might decide to put some mix elements in objects (I mean, beyond the height channels) that I think would lend to upmixing beyond 7.1.4. I deliver that to the renderer as separate sends (with bed channels with those sends omitted). I want to never be reliant on the Dolby software to get back to a mix master. The other thing you could do is run the mix, post renderer, back to the DAW into a 12 channel track to print. Then you can just render it to a standard wavpack file and have a lossless copy that's half the file size and plays in any standard media player.

I'm not sure what to speculate on what went wrong here. I can tell you that the Dolby "downmix engine" has a sort of hard set level it expects or crude limiting kicks in. A hard set threshold for that. -18 to -14 LUFS seems to be the sweet spot. Aside again, I see a lot of -14 LUFS out in the wild. That just straight normalized turned down 4db to make it hit -18 LUFS for streaming services requesting that level. (Less so for the -16 camp and straight delivery to the -14 camp.)

Try this:
Put the final level of the master to -14 LUFS (give or take a db) with -0.2 peaks (or whatever your close to zero preferred peak level is). Run the Dolby renderer to get the .atmos fileset. Encode with the media encoder with dialnorm set to -31. Let me check that result. Or... If you can just render from the DAW sans Dolby, that would be easier and preferred.

I think you've got a full and proper mix after all. I might still have some comments but I think the entire issue here is with the Dolby end. My comments are likely going to be along the lines or backing off the compression and limiting on the drums. Let the dynamic range delivery of this system shine a little so people can crank this up!

 

[mandrix]

I had to turn this up to what normally is earbleed levels for me.
Overall the mix sounds good.
Not a fan of growly vocals, but that's just me. Otherwise the vocals are good, good voice.
There's a "scratchiness" throughout that I'm assuming is the lossy encoding?

 

[LTW]

The wav file sample is free of the lossy sound! The mix balances are different and sound more intentional.

Thanks for the thorough analysis!
Admittedly, I'm still getting my head around how loudness works with Atmos at the Dolby Renderer (it's a lot more straight forward with stereo and even 5.1) but I've gone back over my process to double check everything. I'm pretty damn confident that the mix is correctly setup in Nuendo as I am getting the same loudness reading across three different meters! It's still set at just under -18LUFS but I’ll probably mix to -14 and normalize down to -18 for streaming like you suggested.

I also reimported my exported Atmos ADM back into Nuendo to double check if something was going wrong at export and it looks correct. The loudness matches my mix session. So the ADM is fine.

The issue is with the Dolby Media Encoder and normalization. I encoded the DD+ mp4 at -31 Dialnorm and it makes a big difference to the playback normalization as well as the decoded 12ch Wav. Here's what it looks like now, with the correct loudness.

So it maybe all stems from the dialnorm setting (which was 'No' under Custom Dialnorm before). I had no idea that would affect playback and decoding loudness so much. I'd updated the link at the top with the new version with -31 dialnorm.
https://mega.nz/file/DMF2iSTS#CT06hIygiflTjjjVAindOnBdZIG6hLjtJ7ofHDRszoQ

Dolby Reference Player still has a vastly different loudness interpretation of the DD+ mp4 though, even with DRC off. Plays alot louder. Not sure why. The TrueHD MLP is fine and plays at -18-LUFS (probably because it doesn't employ Dialnorm from what I can tell).

I had to turn this up to what normally is earbleed levels for me.
Overall the mix sounds good.
Not a fan of growly vocals, but that's just me. Otherwise the vocals are good, good voice.
There's a "scratchiness" throughout that I'm assuming is the lossy encoding?

Thanks for listening @mandrix. Yea that scratchiness is from the E-AC3 lossy encoding I think. I will be selling the lossless Atmos mix as well as a dedicated 5.1 mix on IAA.

 

[LTW]

Just found a relevant forum post that explains "dialnorm" and possibly why my mp4 was sounding strange:
https://forums.steinberg.net/t/important-dolby-atmos-and-the-dialogue-level-aka-dialnorm/861632

 

[mandrix]

FWIW, I and some others with a range of AVR's that decode Atmos did a test for a mixer friend. He encoded his tracks (lossless Atmos) at different DialNorm levels.
Some AVR's appeared to ignore the DialNorm settings altogether.
e.g., listening with my Onkyo TX-RZ50 I could tell no significant difference in loudness between any of them.

Since I did not do the encoding, I don't know or more likely don't recall what if anything else he did differently, only that it was reported from individuals that some could tell a volume difference and some could not.

With lossy Atmos (E AC-3) encoding I can definitely hear a difference between different DialNorm settings.

 

[jimfisheye]

OK, now we're getting somewhere!

Even in this lossy copy, I'm hearing more music and less "Battle of the limiters and unexpected levels in the lossy sea"

Flipped around: You can be off by 36db in digital audio and a lot still comes through. That would have been amplified rumble and noise in analog land.

The mix is subdued. Especially the drums. Sounds like it's made to be crammed into stereo. I like the emphasis on fatness instead of ear piercing! I don't want to just suggest turning up high end or straight turning up drum transients (which would equate to turning up high end). But maybe just a little.

I'm trying to be highly critical. The mix is already above average in the industry. But this is a 12 channel mix, right?

I've got -16.4 LUFS integrated, -4.1 peak.
I don't think that's inappropriate. There's still room to let more dynamics in and keep that level and crest factor too. I'd use the headroom up top since it's there but that's just me.

If a different meter reads -18... Not Surprising. I don't mean to open the metering can of worms here! Is that 7.1.4 metering or folded down to 5.1 metering? (Because there isn't agreement on a 7.1.4 standard yet.) As long as the streaming service is happy. Turning the whole thing down 4db or thereabouts is what everybody else is doing too and it's just fine.

 

[mandrix]

Yes. The only thing I can't deal with in lossy encodes (think streaming Atmos) is if there is a large variance in perceived volume across tracks.
I have run into some streamer albums where just one song will suddenly blast at what I assume is -31dB encoded while the rest are pretty tame.
If you have the volume set high and there's one anomalous track, whether too low or too high, then you have to jump for the remote/volume control.

Just something to be aware of.

 

[LTW]

I've got -16.4 LUFS integrated, -4.1 peak.
I don't think that's inappropriate. There's still room to let more dynamics in and keep that level and crest factor too. I'd use the headroom up top since it's there but that's just me.

If a different meter reads -18... Not Surprising. I don't mean to open the metering can of worms here! Is that 7.1.4 metering or folded down to 5.1 metering? (Because there isn't agreement on a 7.1.4 standard yet.) As long as the streaming service is happy. Turning the whole thing down 4db or thereabouts is what everybody else is doing too and it's just fine.

I believe Nuendo uses 5.1 for it's internal Atmos Renderer metering (https://forums.steinberg.net/t/how-...th-dolby-atmos-renderer-supervision/811463/14). I'm not sure what the streamers will be expecting though. Might have to submit a few iterations to get it right.

I did a full playthrough to get -18LUFS. Was -16.4 from a section or full song? -4.1 True Peak matches though. I wish there was a quick full song loudness analyzer. An 11min playthrough takes a while...

The mix is subdued. Especially the drums. Sounds like it's made to be crammed into stereo. I like the emphasis on fatness instead of ear piercing! I don't want to just suggest turning up high end or straight turning up drum transients (which would equate to turning up high end). But maybe just a little.

This is the kind of feedback I was looking for so it's welcome. I'm wary of creating bright fatiguing mixes so I probably tend to pull back on the high end. Studio monitors tend to be quite detailed as well (and I don't have the most high end ones) so it gets pulled back even more. I could look to adding more bright snap to the drums though.
I miss a clear mastering stage with Atmos where things like overall frequency balance is taken care of. And the Atmos Album Assembler is not compatible with PC! Which really sucks. It would be so useful for overall tweaks like this.

The punchy cohesion of rock/metal stereo mixes is tricky to achieve with immersive audio so the safe route is to keep the 'meat' in the L and R speakers. Possible why it could sound 'crammed into stereo'? Dry drums, bass and rhythm guitar are almost stereo. Spreading them out can break things apart. That being said, I pretty much ask myself: W.W.S.W.D. ('what would Steven Wilson do?') . I've studied his Porcupine Tree and solo mixes closely for techniques and tried to replicate the way he spreads his drums and rhythm gats around the sound field. He records WAY MORE rhythm guitar tracks than me though so I've just got 4 (2 really but using tricks to double it for the surrounds). I also use his center speaker vocal treatment that I know not everyone likes.

It was interesting to see that in the recent Atmos mixes for Leprous (kind of the same genre as Look To Windward)? ALL of the guitar, bass and drums are in L and R. The mixes still sound real nice, will lots of synth and vocals everywhere else. But it's an extreme example of keeping the heart of the mix stuck to the front.

I'm trying to be highly critical. The mix is already above average in the industry. But this is a 12 channel mix, right?

That's nice to hear
It's mixed on 7.1.4 but there's enough swirling around to fill larger systems, I'm sure.

Yes. The only thing I can't deal with in lossy encodes (think streaming Atmos) is if there is a large variance in perceived volume across tracks.
I have run into some streamer albums where just one song will suddenly blast at what I assume is -31dB encoded while the rest are pretty tame.
If you have the volume set high and there's one anomalous track, whether too low or too high, then you have to jump for the remote/volume control.

Just something to be aware of.

Oh this is so annoying. Crowded Houses' Gravity Stairs had wildly different levels for some songs. I think it could've been because the singles were released first with different normalization metadata and the rest of the album wasn't matched to them.

 

[jimfisheye]

I meant something like "This is the masterclass mix, right?" in defense of being critical.

The cohesion factor is certainly not lacking! Maybe experiment with adding back in some of the transient dynamics in the drums. Maybe a few other elements. Just doing that would brighten them up without static eq. Think of it like the parts you can only hear when close up to a sound source. Right now you have the zoomed out view that is the only choice in stereo.

Again, I really agree with erring on the side of not too bright! You get high marks as is over mixes you can't turn up because they're piercing.

 

[GOS]

This thread is awesome. I love how quality feedback is being given and being utilized to better the final product. So cool.