AVS Forum banner
41 - 60 of 183 Posts

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #41 ·
but that is compensated for by raising the level trim of the subwoofer output so that it restores the original bass level for each channel AND gives an additional 10dB increase to the LFE channel.
That's not entirely accurate or safe to say.

The LFE is already 10db hot prior to mixing. That remains unchanged.

But more importantly...
If a user cranks it another 10-15db to restore the lost dynamics, and then say say EOT Intro comes along, the LFE was already near clipping (10-30hz), prior to any bass redirection. The center also has bass near clipping (10-30hz) in that movie.

The user's LFE is now 12-18db beyond fully-clipped, and the subwoofer is toast, it has no chance of survival. [and that's exactly what happened to many! ;)]

You cannot restore the dynamics once it is lost (at least, not without huge risk.)

At the time that movie came out I had everything set to all-large, and was vastly underpowered and even then... that movie almost killed $15k worth of subs in my system (subs that can't be replaced). Subs that had 80mm's of excursion and barely made it out alive!
The rest of that movie had like "no bass" because after I had turned it down to "regular levels" the bass never went the hot again for the rest of the movie! (or any movie that has ever come out since! :D)

Let's just say: I now have advanced dynamic-limiters and soft-limiters and hard-limiters in place to prevent that from ever coming close to happening again (especially now that I'm not underpowered.)
I also NEVER crank any trims above 0db, to avoid source clipping for that very reason.
Without that, given the amps I have now, nothing would have survived. :eek:
 

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #42 · (Edited)
When tested by Dolby (using actual program material) for their bass management, it turned out that combining 5-7 channels of filtered bass needed around 5dB of attenuation to keep from clipping. From what I remember, the levels are lowered an additional 10dB, then the LFE channel is added (also 5dB attenuated). When the subwoofer output trim level is raised 15dB, that restores the initial level of the filtered bass and makes the LFE channel 10dB louder. Dynamic range should be back to were it originally was. @Roger Dressler can probably describe it better than I can, since he worked on it.
Hmm... I hope that's not the case. As that would cause clipping in some of the heavy hitter movies.
It is true that it increases the level up (output voltage) and thus restores the actual in-room SPL (at the risk of further clipping by that amount...)
It doesn't restore the lost dynamics though, that is not possible. Once the signal is degraded, it is lost forever.
i.e. The sides, rears and heights will sound like smalls and the subs will be playing clipping.

It is "possible" the disc could be mastered with that in mind to avoid said clipping and bring it to proper levels for all-small systems; but as a user, I wouldn't count on it...

That might work for many movies, especially comedies and chick flicks, but certainly not all, nor from an ideal math perceptive either.

I know why they might do that in the algorithm though (to prevent the users from complaining about weak bass.)

All the more reason to never enable bass-redirection IMO.
I'll just let the bass/full-range signal come out of the objects/speakers, as mastered on the disc... ;)
 

·
Registered
Joined
·
30,632 Posts
The LFE is already 10db hot prior to mixing.
During the mix, the LFE channel is monitored 10dB louder than the other channels, but the signal itself is not 10dB hotter (otherwise it would clip OR the other channels would have to be recorded 10dB lower). In the soundtrack, the LFE channel is the same level as all the other channels; it just gets a 10dB level boost during playback.
You cannot restore the dynamics once it is lost (at least, not without huge risk.)
Assuming dynamic range was lost. Movie soundtracks are 24-bit, which means they have a dynamic range of 144dB. Digitally attenuating the level by 15dB will affect inaudible content 129dB below full scale and quieter. Is going from 144dB of dynamic range to 129dB dynamic range the 'loss of dynamics' you're talking about?
Hmm... I hope that's not the case. As that would cause clipping in some of the heavy hitter movies.
I'm not taking issue with your calculations, just pointing out what happened when tested with program material.
 

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #44 · (Edited)
During the mix, the LFE channel is monitored 10dB louder than the other channels, but the signal itself is not 10dB hotter (otherwise it would clip OR the other channels would have to be recorded 10dB lower). In the soundtrack, the LFE channel is the same level as all the other channels; it just gets a 10dB level boost during playback.
I guess that makes sense, so it is the SP's/AVR's that are boosting it.
I'll have to do some deeper RTA analysis on a few movies to see how my SP is outputting it.
I don't have a software-based Dolby Atmos file decoder, so I'll just have to use the pre-outs with trims at 0db and EQ off (which is how I always run it).

Assuming dynamic range was lost. Movie soundtracks are 24-bit, which means they have a dynamic range of 144dB. Digitally attenuating the level by 15dB will affect inaudible content 129dB below full scale and quieter. Is going from 144dB of dynamic range to 129dB dynamic range the 'loss of dynamics' you're talking about? I'm not taking issue with your calculations, just pointing out what happened when tested with program material.
I'm referring to the relativistic db losses, from the bass-management itself.
The signal before and after remains 24 bit (internally they could even be DSP'ing in 64bit), but the relativistic dynamics are still lost. Even if all the heights and surrounds have tonnes of bass in it (in-phase), you are still (theoretically) losing dynamics or risk clipping in the restoration.

To avoid such clipping they could of course use intelligent normalization, rather than blind static amplification. Which is entirely possible and would do a better job, but I'm not sure if that is the case with any/all gear, and if so... Is that a Dolby firmware thing or left up to the manufacturers to ignore?
 

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #45 ·
I told you it was there! That why I was hoping you could test and confirm as I didn’t have very much atmos content on hand to play with 🙂
I guess at the end of the day there will still be users who will run everything as smalls + bass management.

and then there will be users like me, who try to run everything as all-large and play what is actual on the disc, because if the movie-director / studio-people wanted "all bass to be in the LFE, optimally", then it already would be... ;) :D
 

·
Registered
Joined
·
12,412 Posts
I guess that makes sense, so it is the SP's/AVR's that are boosting it.
Yes, all the bass management summation math is performed to prevent clipping. Then the analog audio is gain adjusted as necessary to drive the subwoofer.

I'm referring to the relativistic db losses, from the bass-management itself.
The signal before and after remains 24 bit (internally they could even be DSP'ing in 64bit), but the relativistic dynamics are still lost. Even if all the heights and surrounds have tonnes of bass in it (in-phase), you are still (theoretically) losing dynamics or risk clipping in the restoration.
There is nowhere in the DSP chain where clipping takes place. The first point of clipping, in a properly built system, would be the power amplifiers.

To avoid such clipping they could of course use intelligent normalization, rather than blind static amplification. Which is entirely possible and would do a better job, but I'm not sure if that is the case with any/all gear, and if so... Is that a Dolby firmware thing or left up to the manufacturers to ignore?
The degree of headroom scaling may change based on the source, but other than noise floor, nothing changes.
 

·
Registered
Joined
·
30,632 Posts
Is that a Dolby firmware thing or left up to the manufacturers to ignore?
Manufacturers are free to implement their own bass management (Lexicon used to do that), but it is much easier to use the Dolby bass management (well thought out and comes loaded on the DSP chips anyway).
 

·
Registered
Joined
·
376 Posts
This is why I ask you guys! Thank you!

So what you're saying is, go full @BassThatHz on every channel AND subwoofers :D
Woulda shoulda coulda. In the end as many have pointed out, the bass will get added to the sub output based on the individual speaker capability. I just ordered 4 volt-6s but I got them ported because I would rather have them go down a ways, but at some point it is what it is. I'm not going to hang a 7 cu ft box from the ceiling! It'll be good enough. You can spend a lot of time and money chasing something that may not as yo to a hill of beans!

Sent from my Moto G (5) Plus using Tapatalk
 

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #50 · (Edited)
I recorded the 11.1 output of the MI:Fallout helicopter crash scene. Here are the results:



^Center
-3.57db Peak
-20.37db Avg
27db Dynamic Range
135hz was loudest note


^Height Back Left
-5.84db Peak
-85.34db Avg
88db Dynamic Range
135hz was loudest note. Pretty solid bass to 20hz up there.
I think the opening nuke/rain scene and skydive scene and helicopter chase scene it goes a bit deeper, closer to maybe single digits, not sure...



^LFE
-5.19db Peak
-30.48db Avg
77db Dynamic Range
31hz was loudest note

I'm surprised that the heights are that dynamic, I guess it makes sense since they go from basically muted to full amplitude. That is a HUGE swing in power though, 631million times increase in power.
Even the 77db dynamic range of the LFE is pretty large, 50million times increase in power.

Now I'll repeat the test for EOT Intro and see what we get, which I don't have in Dolby Atmos, I believe it is just 7.1

I googled some online calculators and they say that for 12 sources in works out to 11db total loss, not 15db so I guess we were off a bit.
This is good to know because I have created a custom 11.1 Dolby Headphone downmix, my SP was only doing a 5.1 TrueHD downmix.
 

Attachments

·
Registered
Joined
·
12,412 Posts
I recorded the 11.1 output of the MI:Fallout helicopter crash scene.
I believe you are using analog outputs. How are you calibrating these to the measurement system?

I googled some online calculators and they say that for 12 sources in works out to 11db total loss, not 15db so I guess we were off a bit.
What exactly are you calculating? It looks like 10log12,which is 10.8. The theoretical sum of 11 main channels plus LFE comes out to 13 dB higher than LFE at 0 dBFS.
 

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #52 · (Edited)
EOT Intro.

May god have mercy on your LFE and Center. :eek: Like a noob air dropped into a battlefield with no bullets. :p

My SP upmixes it to fake-mos.

The other channels are so uneventful that they aren't even worth scanning, at least in the intro...


^Center
-2.70db Peak
-32.9db Avg
97.66db Dynamic Range


^LFE
-0.92db Peak
-50.68db Avg
92.32db Dynamic Range

The whole 1st battle:

Clearly the center is the most used

^Center

^LFE

^Right Front

Zooming in on the Intro LFE specifically:

36db hotter than the battlefield scene, which is already 98db louder than the noisefloor of my analysis rig.
Basically: as close to going from all-0's to all-1's as you'll ever see, the full 24bits of subwoofer destruction! :eek:

Now the question is, what happens to the LFE when we set things to small?
 

Attachments

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #53 · (Edited)
I believe you are using analog outputs. How are you calibrating these to the measurement system?
Good question.

I'm connecting my 7705 to a Motu 24AI via XLR. Both set to 0db. AKM DAC to Sabre Reference DAC.

Feeding the 7705 via WASAPI HDMI PCM at 0dbFS, I get -4.787db too cold. So given that EOT was able to hit louder than that, that means that either the 7705 is holding back on PCM signals vs Atmos, or Windows is reducing it prior to the HDMI output (or input.)

The waveform is clean and unclipped, even at 10hz.

and repeating it via the Motu AVB digital loopback, I'm down 0.132db.

Which tells me that it isn't a windows or motu driver or motu digital system problem,
it's the 7705's analog output that is holding back on the voltage, changing levels from: PCM vs Atmos/TrueHD.
I tried pure-direct and normal mode, and it's the same way.
Perhaps the 7705's LFE output is naturally 4-5db hotter than the LR. The 7705 doesn't not support ASIO for 7.1 PCM.

I don't have an Atmos file with a 0dbFS sinewave so I can't compare the decoder differences. The EOT intro is likely the hottest dolby test signal I have, I couldn't imagine a louder track existing...

Here are several RTA's of that movie done by others, it's definitely a bassy intro:
http://data-bass.ipbhost.com/topic/304-edge-of-tomorrow-discussion-poll-closed/
data-bass.ipbhost.com/topic/425-analyzing-waveforms-of-heavy-hitting-movies/
https://www.avsforum.com/forum/113-...-audio-subwoofer-thread-963.html#post52666129
 

Attachments

·
Registered
Joined
·
9,729 Posts
Summing redirected bass does not rob the lfe channel of dynamic range. If your subwoofer system isnt capable, it may cause you to experience dynamic compression, but thats not in the signal path, thats a lack of woofage or wattage. This was maybe a big concern 10+ years ago, but anybody whos legitimately into the hobby nowadays has multiple 18s or larger.

If one has capable speakers, the only issue with running everything full range is that every speaker position is not going to have good bass frequency response at your seat. Thats when isolating each speaker. Now have all 13 channels in 13 different spots play 13 different full range signals and the bass will be doing all sorts of ugly things at your seat.

With a sub system, you can typically achieve better placement and then EQ the frequency response to your seat. Now ALL the bass from ANY channel makes it to you more intact.
 

·
Registered
Joined
·
12,004 Posts
Discussion Starter · #55 · (Edited)
Interesting results.

I set all the speakers to small 80hz.



The center is so hot in this movie that it is STILL acting like a 40hz 6db HPF. LOL!


and the LFE was so hot now that I had to set the Motu input to -2db just to keep it unclipped. So take the difference of the peak values, and add another 2db to it! 1.08db hotter! But still unclipped once I reduced the input sensitivity 2db.

^bass managed LFE, 80hz all small system

1.08db hotter Peak!
-28.79db Avg, 21.89db hotter!
Dynamic range: 56.48, lost 35.84db of dynamic range!

That's because the LFE used to be dead silent when there wasn't a specific LFE effect, and now it has rumbles from the other 11 speakers coming out of it, constantly!
and that IS a noticeable\audible change in subwoofer tonality!

i.e. used to be dead silent and now has road noise from the LR, and vocal noise from the center (males), and helicopter swash sounds from the heights, and rain noise from the surrounds etc etc etc. I can't say that it sounds good or better!!! :eek:

and my large speakers are now sounding like small bose modules, and that is causing an SPL drop because the LFE is nearly at the same Peak SPL as before, but the other 11 speakers are now quieter (neutered below 80hz.)
 

Attachments

·
Registered
Joined
·
12,412 Posts
I'm connecting my 7705 to a Motu 24AI via XLR. Both set to 0db.
Does that mean 0 dB gain? I'm not sure that gain matters. The idea would be to produce a known digital signal level in the 7705 and convert its analog output to that same level in the Adobe Audition measurement. Adjust the gain until they match.

You do not need full scale signals to calibrate. If you have REW, it can generate sine waves at whatever level you like, -20 dBFS or even 0 dBFS. That's the same level you should see in the meter.*

*Just note that measurement tools offer two different references for 0 dBFS, one being full scale sine wave, the other being full sale square wave. They both have the same digital peaks (as read by "true peak" meters), but the square wave reads 3 dB higher on RMS meters (which is what we usually use for voltage/loudness measurement). In Adobe Audition, this option is present in the analyze window.
In REW -- it's in Preferences / View. There's a checkbox for "full scale sine rms is 0 dBFS". If you check it, full scale square waves can read +3 dBFS. For these tests we'd leave it not checked (square = 0 dBFS mode) so that 0 dBFS is only reached when audio is smashed into square waves. :eek:

Feeding the 7705 via WASAPI HDMI PCM at 0dbFS, I get -4.787db too cold. So given that EOT was able to hit louder than that, that means that either the 7705 is holding back on PCM signals vs Atmos, or Windows is reducing it prior to the HDMI output (or input.)
Or maybe the meter calibration is not accurate. Hard to say from here.
 
  • Like
Reactions: BassThatHz

·
Registered
Joined
·
353 Posts
..... and then say EOT Intro comes along.....:
I've really enjoyed reading this thread, but forgive my naivety, what soundtrack or what is being referred to with "EOT".

I'm tempted to switch my AVR settings back to large for my bed speakers in a 7.2.4 system. To each of the channels in the bed, I have connected each their own subwoofer, in effect I call it a third "active" driver as I have them placed either under or right next to each tower. (even have on in use with the center channel, inside the cabinet). Currently I have my crossover set to 40 hz. Even so at 40 hz and below as per this discussion, there's still a ton of information going to the LFE channel plus the LFE.
I'm going to at least play around this weekend and see what my ears tell me. In any case, fun topic for sure.
 

·
Registered
Joined
·
2,067 Posts
How are you doing the crossovers from the bed channels to the subs? What speakers and subs are you using?

Sent from my SM-N920V using Tapatalk
 
41 - 60 of 183 Posts
Top