How to decode 6-track APTX-100 (cinema DTS) with the correct channel levels

[borisanddoris]

Surrounds universally need to be lowered -3dB since the B-chain always set them (in a properly aligned room) to 82dB vs. 85dB at home. This still holds true for a theatrical mix that is ported to the home today. If you have a full surround array, you could probably get away with 82dB.

The LFE is the more mysterious one for sure. I too have found the +3dB to work well for pre 1999 films except Jurassic Park. I’d like to get a hold of some other earlier DTS titles like Heart and Souls or Hard Target to check them out.

All the DTS6 install guides I’ve seen date well after 1993 and would be interested to see if there’s something different, even slightly, in older documentation. Albeit I’ve spoken to many folks who ran Jurassic Park and the 88dB SPL for LFE was a thing with no special considerations to the contrary.

[2001CluelessDianogas]

So, here's what the measurements should apparently be:

L, C, and R, measured with an SPL meter, playing internal pink noise or pink noise from DTS Setup Disc or Empirical Disc:
85 dBC

Ls and Rs, measured with an SPL meter, playing internal pink noise or pink noise from DTS Setup Disc or Empirical Disc:
82 dBC (-3 dB compared to screen channels)

Subwoofer, measured with an SPL meter, playing pink noise from DTS Setup Disc or Empirical Disc:
91 dBC (+6 dB compared to screen channels, but pink noise is only ~20-80 Hz because of DTS processing)

Subwoofer, measured with an SPL meter, playing pink noise from the internal pink noise generator of the DTS-6AD:
85 dBC (same as screen channels, for some reason, which I'll get to in a sec)

Right, so this brings me back to one of the things that was baffling to me before: why is the DTS-6AD's internally generated pink noise for the subwoofer apparently 6 dB quieter than the pink noise off a DTS Setup Disc or Empirical Disc?

When @schorman calibrated his DTS-6AD completely flat (no gain adjustments on any channels) then recorded some soundtracks through it from films released in or after 1999 (thus complying with the new SMPTE RP200:1999 subwoofer level standard), it seemed to output the subwoofer channel at a level that matches other releases of theatrical audio on LD and DVD releases of the same films, but the surrounds still needed the -3 dB cut. The foobar2000 plugin, by contrast, has the subwoofer channel 6 dB quieter than the DTS-6AD, so it would seem to be the case that the DTS-6AD automatically handles the +6 dB subwoofer boost for SMPTE RP200:1999 films but the foobar2000 plugin defaults it to 6 dB lower than it should be unless you manually change the setting. Curiously, neither the DTS-6AD nor foobar2000 plugin seem to apply the -3 dB cut to the surrounds: in both cases, the surrounds seem to be 3 dB louder than the LD and DVD releases of the theatrical audio mixes.

Based on the measurements of the DTS-6AD's subwoofer output, I'm wondering if this -6 dB measurement discrepancy (85 dBC for internal pink noise, 91 dBC for pink noise from setup disc) maybe has something to do with the unit already adjusting the subwoofer output by +6 dB when it's decoding off a disc.

Yeah the docs are a confusing mess. And it gets confusing about when they are talking about home the speakers are set, how the data is stored, when this or that is on, whether they are comparing full or partial bandwidth, at what stage, etc. the mess where home theaters set subwoofer to same as regular channels but theaters have the subwoofer dialed up +10dB and how most cinemas have full bandwidth mains and most home cinemas re-route bass from mains to subwoofer and don't use full range mains. For home theaters bass re-routed from mains to sub needs to be played at same level as mains, but sound from LFE channel from movies and properly LFE calibrated music discs needs to be then boosted +10dB at last stage to sub to account for movie theaters having sub that plays the LFE +10dB to mains. If you just dial up home subwoofer +10dB then you get the mess that re-routed bass from mains is way too hot. And most A/V/ receivers have weak and confusing documentation as to what they do, especially if using analog inputs, etc. And same for much PC playback softare and crossover sotware, little of describes what it does, does it boost LFE or not? how does it handle crossed over vs LFE volume differences, etc. More of a mess if you use analog output from soundcard. It is easy to end up with way boosted rerouted bass from mains for music or movies or with way too soft LFE effects or only one or the other correct at one time, etc.

Anyway, the surrounds need to be lowered -3dB by foobar when you decode because home theater designates that you set L,C,R,Ls,Rs to all read the same volume for the same input but in cinemas their Ls and Rs each put out -3dB less volume for the same input than L,C,R so they are calibrated -3dB compared to for home theater. So when listening to CinemaDTS mixes you either play it as is and then set -3dB for each surround or simply leave your home cinema speaker levels alone (safest, since not everyone will know to adjust) and then simply lower the signal -3dB. The cinemas put their Ls and Rs -3dB lower each than L,C,R as an old artifact from when surround was mono and for easy auto compatability if it dropped back to some optical mono surround or whatnot since that gets fed then to Ls and Rs they'd have mono surround volume doubled otherwise. So the Cinema DTS decoder doesn't do any -3dB for surrounds since they don't have to, the cinema surround speakers are already each calibrated -3dB lower unlike for home setups.

[2001CluelessDianogas]

Thanks so much for all of this information. I'm in the process of doing some Cinema DTS syncs at the moment and I'd like to make sure I'm setting the levels correctly. How certain is it that all post-1999 movies need an LFE boost of +6db?

If I apply a +6dB LFE boost to any of the LOTR movies then it results in clipping; +4dB is about as loud as it can go which incidentally puts LFE at the same level as the BD has it.

JPIII can be boosted +6dB without clipping but this makes the LFE +2dB louder than the DTS DVD track (all other channel levels of this track match the theatrical). Again +4dB would match the DVD.

I'm converting with the foobar2000 plugin v0.1.4. Also these movies were released in DTS-ES 6.1 with the additional channel matrixed into the surrounds. I don't know if this makes a difference. I'd be grateful for any insight.

I have not done a DTS-ES 6.1 yet, not sure if that affects things.

If not and for the others:
First when using foobar turn off it's auto LFE creation and all of it's +dB processing for any channel, let it do nothing at all. I seem to recall that you set it to do LFE and adjust dB of anything that the dB amounbt adjusted seems different numbers than when doing it manually so not sure what it is doing under the hood. And use the most recent version.
Take the Ls,Rs and combine the two tracks and then do a lowpass at 80Hz filter on them and then save that out as the LFE and then do a +3dB gain to it (already have a +3dB from combining the signal stored in both Ls AND Rs). For pre-1999 I think forget the +3dB stage. Smetimes you need to do the 80Hz lowpass filter first before combinging them as the upper frequency power might clip stuff, in those cases, 80Hz filter first, then combine the two channels, then +3dB if post start of 1999 release. Actually it's been a while, I forget, I think maybe in the end I did NO +3dB for the early 1999 title actually and decided it maybe just needed nothing done since they had recorded it in studio with LFE speaker already set to account for+3dB and then they stored it once in each channel and combining gives another +3dB. So maybe it is actually low pass, combine and then -3dB for pre-1999 and as is for 1999+? I forget, will check. I know that whatever I did, in audactity, my LFE looked about the same as what Schorman got for AOTC and all I did was either lowpass each and then combine Ls and Rs and then leave them be or at most then do a +3dB at that point and did no +6dB or +9dB and such for sure.

Then undo those actions on the Ls and Rs and then apply -3dB to each and then apply a high pass filter to them at 80Hz.

I was doing 48dB octave passes in Audacity although that seems to remove more stray high frquency from LFE than what the CinemaDTS box does looking at Schorman ouput so that box maybe does something more like 24dB per octave but it also seems to use a different filter method too so it won't be quite the same whatever way.

My software method in Audacity left a bit less instances of clipping than the Schroman DTS box output, neither had many clips at all, but the official box output had a few more. Or perhaps it came from when he resampled it from 44kHz to 48kHz? Not sure if resampling can produce clipping, have to think, maybe. I know applying low and high pass filters can since even though you are removing total energy you can realign some stuff to re-enforce and can phase shift some frequencies relative to others and they can add in new ways and they can also produce peak artifacts near cut frequency, etc. As even his center for one disc had the part that also clipped one mine (apparently clipping in the master!) clip just a bit more, maybe that was from his 44->48Hz resmapling stuff.

To best avoid clipping you may need to play around with the order of steps and balance clipping vs. maintaing max SNR:

For one title (using Audacity and whatever algorithms and methods it uses) reel 1:
I have to first -3dB before high-pass to create surrounds if I don't want any clipping in them.
I can straight add the two and then lowpass and make LFE with no clipping.

for reel 2:
I can just high-pass the surrounds first without clipping and then give them their -3dB.
I can straight add the two and then lowpass and make LFE with no clipping.

for reel 3:
I can just high-pass the surrounds first without clipping and then give them their -3dB. EDIT: nope can only do that for left surround, right surround need to first -3dB and then high-pass to avoid clipping.
I can't just straight add the two surrounds and then lowpass and make the LFE without it clipping. Even doing -3dB first to each surround and adding them still clips. I found I could first low-pass them and then add them as is without clipping though.

etc.

I wonder how the DTS processor does it. For safety does it just first -3dB the surrounds and then high-pass and then raise them back up again 3dB (remember the theater speakers have surrounds set -3dB) and then also add the two -3dB adjusted (EDIT: nope this is not enough to stop clipping for some of them when starting to make the LFE) together to make the LFE and then low pass that and then raise that back up again 3dB? Does it just apply all the stuff and the adding as they are and you end up with some clipping? In which case these manual methods with care may give more distortion free output than the theatrical processor. Maybe it does -3dB to each surround and then high-pass filters them to produce the surrounds. And then it goes back and looks at the original combo surround data and does a low pass filter on them and then adds them together.

[2001CluelessDianogas]

One thing to keep in mind is how you are outputting this stuff and whether your PC is doing the +10dB (+6dB depending how you measure stuff) boost for LFE channel to your sub and your sub at home is normally set +0dB and not +10(+6dB) like in a theater, in a theater the sub is the LFE while in most home theaters the sub is the LFE signal (which needs to be +10dB) and the re-reouted deep stuff from all other channels (which needs to be +0dB). Etc.

How does your A/V receiver handle .1 input from analog vs PCM digital inputs vs DTS/DD etc.

[maksnew]

Perhaps I just didn’t see this in various comments, but most likely this is not the case, when changing the sample from 44.1 to 48kHz in most modern editors, a frequency cut occurs and quality is lost by 20k in the frequency range, which already becomes acceptable only for DTS but not as not for DTS HD, I certainly found a way out, but the result of the deterioration was very unexpected for me.

[2001CluelessDianogas]

Actually I think that all you do is sum the two 80Hz low pass Ls,Rs together and that gives you the LFE for 1999+ discs and for the pre-1999 you actually need to then do a -3dB on them.
I don't think you do a +3dB add or a +6dB to the LFE (at least not when re-constructing the LFE manually yourself).

Basically 1999+ assumes same setup as for home where receiver or playback software or whatnot is adding a +10dB (+6dB with the other measuring method) and so 91dB LFE, 85dB L,C,R but home has Ls,Rs at 85dB vs. 82dB for theatrical so you do need to do a -3dB to Ls and Rs on your CinemaDTS decoding. But the LFE is already as it should be and your playback software/receiver should be giving it an autamtic +10dB (+6dB boost). If you use analog out you have to use software that allow proper LFE vs crossed over bass from other channels stuff properly, you can do it with JRiver by setting up filters and boosts and lowerings. For DTS/DD over digital receivers automatically boost the LFE. If you convert your Cinema DTS conversion to DTS MA or TrueHD the receiver should properly auto boost the LFE. If you leave it as PCM not sure maybe some receivers don't add boost to .1 and the you'd need to see if their is an option (probably should be since many discs used to be all PCM audio) but if not then set up software to control it.

Anyway basically when manually creating all channels, leave everything as is but -3dB to Ls,Rs and then LEAVE LFE as is, no boost! for 1999+ discs and actually give is a -3dB (minus) boost for pre-1999 discs. This does make JP LFE seem lower than people expect but I think most to all home versions have actually been wildly upping the LFE on it. (although actually I was using a later JP cinemaDTSdisc and need to verify it's actually still pre-1999 and mastered to 88dB LFE and not 91dB, have to compare to original initial JP discs and see if it is same levels, maybe the later discs refrmatted to new .aud system are 1999+ and need no -3dB on LFE, they have THX trailers up to TEX 1 but that is still 1996 so the disc is probably not 1999+ but will check, maybe they just felt more ecent TRHX were not a good mix for this film).

And yeah I just keep it 44kHz why resample to 48kHz and add some more artifacts? I also leave it a 24fps timing for same reason. It fits 35mm scans then and if you want to mix it with video from a commercial disc and it's not one of the relatively few done at 24fps then just tell the program to treat the video like it is 24fps and combine the audio, no alteration of anything then other than video playback rate metadata.

[2001CluelessDianogas]

Oddly the decoder plug-in, all versions beyond the very original 1.1, crash out, give negative starting point and don't work or play back garbage when trying to decode the original JP Cinema DTS discs although the later re-formatted updated discs decode the same as anything else with all the versions. Only original 1.1 version seems to work with the .r1 .r2 .r3 .r4 etc. files on the original less than a handful of Cinema DTS discs. Still need to verify if they potentially used a different LFE than either 88 or 91.

[stwd4nder2]

I don't remember if it was this forum or another but someone out there compared the LFE levels with PCM recordings of discs being decoded by theatrical DTS hardware, and the +3/6db rule seems to be accurate.
Without significant evidence pointing out why that shouldn't be the case I'll continue to go with that when doing my syncs.

[2001CluelessDianogas]

I don't think they did that correctly or maybe they were comparing to what some particular version of the plug-in did when it created the LFE channel itself rather than constructing it manually or comparing to boosted home versions or maybe it's just that it was based on manually creating the LFE but doing that after having already done a -3dB to each surround rather than before in which case you would then need to undo that for the LFE and add back +6dB (or just +3dB for pre-1999 discs since they are encoded already +3dB too hot).

The Cinema DTS specs simply say that F/C/R should be full range speakers (20Hz-20kHz) and Ls/Rs should be (80Hz-20kHz) capable and that subwoofers (which get only LFE signal) should be set to 88dB for pre-1999 discs or to the standard 91dB for post 1999 discs, L/C/R to 85dB and Ls/Rs to 82dB (so that when playing legacy mono surround getting fed to each channel it delivers to an 85dB standard) and that is what they mixed for.

Home setups use L,C,R 85dB, Ls/Rs also at 85dB (rather than 82dB) and subwoofer usually also set to same as L/C/R/Ls/Rs and not raised +10dB full band equivalent since music expects subwoofer to be same level as other channels and not have any boost. However, the LFE channel for movies should run at +10dB (full band equivalent, but measures at +6dB when sending a restricted 20Hz-80Hz test signal using C weight probe measuring).

So home theater processors when they detect cinema multi-channel like DTS/DD/TrueHD etc. boost the .1 LFE channel up before sending the signals out the analog speaker connection wires (for some multi-channel music encodes I think this needs to be turned off) but since many home speakers have deep bass crossed over for L/C/R and especially Ls/Rs that crossed over bass from those channels needs to stay at regular level so they don't boost that but just add analog boosted .1 LFE to the unboosted combined crossed over bass.

I'm not sure what they tend to default to when they detect multi-channel PCM, maybe you need to make sure the LFE gets a boost one way or another yourself with some setups not sure at the moment. I actually run analog out to my receiver from a high quality soundcard with replaced top line opamps.

When doing analog out to receiver you generally have to account for everything manually yourself. A lot of playback movie software is very vague on what it decides to do in this case. Mostly I think they do nothing and then it's seemingly not possible to get the LFE boosted without also incorrectly boosting crossed over bass. I think Windows Media Center actually seems to lower all channels -10dB aside from LFE channel since it's volume seems to play files back about that much lower than programs that appear to do nothing in this scenario. If you then set audio card software to do bass crossover hopefully it does that at last step, think it does and then it's all OK. Not quite sure about this all regardng WMC though. Anyway, something like JRiver Studio lets you apply all sorts of custom audio options under parametric equalizer section of DSP studio and there you could do:

-10dB (Sub,Left,Right,Center,SL,SR,RL,RR) (-6dB when doing music and not movies) (lowering everything to not overload and clip when crossing over bass and also raising LFE channel up, probably this is safe enough value)
Copy Left to U1
Copy Right to U2
Low-pass at 80Hz (U1,U2) (use whatever is ideal for your particular L,R speakers and later for C and surrounds in terms of the Hz for all of these low and high pass steps; perhaps try these all at a steep 48dB/octave, not quite sure what would be ideal though, maybe some should be less steep)
High-pass at 80Hz (Left,Right)
Add U1 to Sub (-10dB) (adding crossed over bass at -10dB relative to LFE channel and once sub gets a +10dB at the end it will end up balanced again with regular channels and LFE boosted 10dB above them)
Add U2 to Sub (-10dB)
Copy Center to U1
Low-pass at 100Hz (U1)
High-pass at 100Hz (Center)
Add U1 to Sub (-10dB)
Copy SL to U1
Copy SR to U2
Low-pass at 120Hz (U1,U2)
High-pass at 120Hz (SL,SR)
Add U1 to Sub (-10dB)
Add U2 to Sub (-10dB)
Low-pass at 120Hz (Sub)
(if you want you could also high-pass sub at some low Hz like 30Hz or 20Hz or something to add some sub protection if you sub doesn't go super low)
+10dB (Sub) so in the end all regular channels are at -10dB as is crossed over bass and LFE is at 0dB
(for music you didn't really need to add the crossed stuff over at -10dB and then raise it back here by +10dB, could just uncheck this box and add at 0dB but then you need to keep entering new numbers when swithcing movies to music or make a bunch of different add commands and swap check marks for a bunch of things instead of just having to to swap between -10dB and -6dB commands that the top of the list, jsut a single check box swap)

but anyway back to the topic at hand since home theaters are set to have LFE channel running at the 91dB figure rather than the 88dB figure which pre-1999 Cinema DTS discs were mastered for but the same as for which 1999+ discs were mastered for that would certainly seem to me to imply you do +0dB for 1999+ discs and -3dB for pre-1999 discs (not sure when the transition in 1999 occured off-hand) rather than +3dB and +6dB

I'm not quite sure with respect to what Schorman was adding +3dB and +6dB, maybe that was to auto LFE creation from some version of the plug-in that set things 6dB too low (in which case it would be correct but only if you let the plug-in create the LFE for you and probably only with certain versions of the plugin)?

Maybe it's a mistake and trying to account for the raised LFE channel relative to other channels that its taken care of during playback decoding and should not be applied in the audio data itself?

Maybe it was compared to extracting LFE from only a single surround and not both (although I'd think you need to do a +0dB and +3dB then)?

Maybe the home video discs he was comparing to were somewhere in the +3dB or +6dB too hot (souped up for drama or because maybe LFE on many home discs has a lot of the lower bass from surrounds already pre-mixed into LFE since most home surrounds are not 80Hz capable much less 20Hz so the home LFE comes out louder since it has more bass from non .1 stuff already pre-mixed into it?) and not a proper reference?

Maybe he was manually creating LFE but doing it from surround channels that he had already processed to each be lowered -3dB (if it was created manually this way then you would need to give it a +3dB or +6dB boost; if you first combined the surround's sub 80Hz info together before you did any -3dB to surrounds then you'd not want to do the +3dB or +6dB boost though)?

Maybe I've made a mistake but I don't seem to see it.

[stwd4nder2]

You make a solid argument. But if we assume that an audio processor will treat 5.1 PCM the same way as 5.1 DTS/TrueHD (which would make sense, you wouldn't want your calibration to be off because of some variable like that) then there's still the very convincing matter of the LFE matching blu-rays that more-or-less port the theatrical 5.1 when using the +3/6db rule. The Nolan UHDs are good examples, but I've encountered various others throughout the years. There's also the very subjective argument that I've been listening to tracks that have been editing this way for awhile now and I think they sound right, but take that with a grain of salt.

We could always objectively test weather or not LFE output is the same for bitstream/PCM with an SPL meter. If I can find mine I'll break it out and test.