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

[pipefan413]

Before I start, this is not about the home video version of DTS, which is actually called DCA (for DTS Coherent Acoustics). It's specifically about the DTS cinema codec, which is APTX-100 (or APT-X100, depending on where you look). Just in case anybody gets confused by them both being "DTS".

I'm writing this mostly to solidify some stuff that I only even thought about thanks to @schorman noticing that the levels of the surround and subwoofer channels from the output from the foobar2000 APTX-100 plugin did not match those from an actual DTS-6AD cinema processor that he has fairly recently tested. A huge amount of credit therefore has to go to him here, since I wouldn't even be looking at it otherwise, and the information gleaned from testing the DTS-6AD was instrumental to a lot of what I'm about to say, including the actual numbers for the foobar2000 plugin (which I just verified after he worked them out, rather than realising them on my own). I'm just compiling the info here in case anybody else might benefit from it.

This will be a very long post (surprise surprise). In case you don't really care about ther "why" and just want to know the "how", I'll give you the TL;DR up front.

The foobar2000 APTX-100 plugin does not calibrate the relative levels of its output in accordance with the DTS cinema standards. To correct for this, you need to reduce the gain of the surround channels by 3 dB, and increase the gain of the subwoofer channel by either 3 dB or 6 dB depending on the original release date of the particular film you're decoding. Films released on or after the 1st of January 1999 need the subwoofer level increased by 6 dB, and films released before then only need it increased by 3 dB.

If you're using the current version of the plugin, you can correct the subwoofer level natively inside foobar2000 (but not the surrounds). Go to File\Preferences\Advanced\Tools and you'll see the settings for the APTX-100 plugin. Change "LFE level" to your 3 dB or 6 dB, depending on the film's release date.

For the surrounds, you could do this one of two ways. The first way would be to reduce the surrounds directly by applying a manual gain adjustment of -3 dB to each of them and dithering appropriately, which in my opinion should always involve auto-blanking so you don't dither silence (this is particularly important if you're going to encode to DTS-HD MA for delivery because it won't decode properly if the file doesn't have clean silence at the start of all channels). The other way would be to just have your encoder reduce the surrounds by 3 dB, which you can do in the DTS Suite and probably many other encoders very easily by just flipping a switch!

That said, I'll move onto the "why", for the curious...

APTX-100 does not actually store every channel at the "correct" level per se, but rather in accordance with a set of guidelines for calibrating them to your particular theatre. This is not really taken into account by the APTX-100 Winamp or foobar2000 decoder plugins, which (by default at least) just decode it as it is, leaving the levels as they're stored. This is not necessarily inherently "wrong" as such, but for things like fan projects that utilise cinema DTS audio, you're not quite reproducing the sound as it was meant to be reproduced if you don't correct for this calibration at delivery. All I'm going to explain here is how a cinema screening room is calibrated for DTS and how this relates to decoding it for alternative purposes, but it's important to note that a living room is not going to be kitted out the same way as a cinema (generally speaking), nor is it likely to be the same size, so these calibrations won't necessarily produce "appropriate" results for home use. They are, however, correct to how it was done in the cinema.

Basically everything I'm about to say from this point onward comes from either the November 1999 version of the DTS post production guide (http://www.film-tech.com/warehouse/manua...STPROD.pdf) or the operation manual for the DTS-6AD cinema processor (http://www.film-tech.com/warehouse/manuals/DTS6AD.pdf). I'll quote both, but you might find other things in there interesting as well.

First off, here are the guidelines for correctly calibrating a cinema for 6-track DTS from 1999 onwards:

5.3.1 SPL ADJUSTMENT

(...)

• Repeat procedure for all channels, set to the following SPLs:
  
  Left, Center, Right 85 dBC
  Left & Right Surrounds: 82 dBC
  Subwoofer: * 85 dBC
  
  * NOTE: The above levels are for use with the internal pink noise generator. When playing back a DTS Set-up or Empirical Disc, the subwoofer level should be 91 dBC. This setting conforms the subwoofer level to the SMPTE RP200 standard of having +10dB in-band gain in relation to the screen channels.
  

(...)

5.3.4. DTS SUBWOOFER LEVEL

(...)

• If a RTA is not available, the DTS subwoofer can be measured using a SPL meter.  The SPLmeter must be set for “C” weighting and “slow.” It should read about 91 dBC when the (DTS Setup Disc) subwoofer pink noise is playing in theater. Take measurements in different areas ofthe auditorium to prevent subwoofer from being adjusted too loud.
  

The orange text is me adding emphasis, but all other emphasis (bold, italics) are as they appear in the original document. I'll explain why I've highlighted the bits I have in a moment.

What's essentially happening here is that the front channels - L, C, and R - are being adjusted until they all measure at the same SPL (sound pressure level), and the surrounds and subwoofer are being adjusted to comparatively different levels. The correct calibration for L/C/R is to use pink noise to set them to 85 dBC SPL, then the surrounds are set 3 dB quieter at 82 dBC, and the subwoofer 6 dB higher at 91 dBC. Note that I'm saying "subwoofer" here because it's not necessarily the same as "LFE". See, APTX-100 doesn't actually have a dedicated LFE-only channel: it stores the low-end audio intended for the subwoofer(s) in the surround channels, then pulls it back out at playback, similar to 70 mm Todd-AO/Dolby. Here's some info about this from the manual for the DTS-6AD cinema processor:

5.3.3. DTS DIGITAL SURROUND SIGNALS

With the RTA connected, notice a dramatic roll-off at 80Hz. DTS derives the digital subwoofer by filtering out the surround signals from 80Hz and below. This is normal when in the DTS digital format and surround pink noise is playing in the theater.

Anyway, that's what calibration looked like in 1999. For anything before that, you need to be aware of the following change that happened that year:

Subject: SMPTE RP200 changes DTS subwoofer level (REVISED), August 1999

At a SMPTE meeting held in late 1998, all three digital sound companies agreed to comply with the recommended practice for subwoofer level, SMPTE RP200. The subwoofer level has changed to enable the use of one master recording when transferring to all three digital sound processes and to provide playback consistency in theaters. As of January 1, 1999, the recorded subwoofer level on 6-track masters has been lowered to the SMPTE recommended level of 10dB in-band gain (as compared to the screen channels).

To comply with this new standard, all DTS films released in North America after January 1, 1999 have been transferred into the DTS digital process with subwoofer at 10dB in-band gain. The SMPTE RP200 logo is clearly visible on the discs of these films.

Because the recorded subwoofer level has been lowered, the DTS theater subwoofer playback should be increased. Increasing the DTS subwoofer level compensates for the new lowered recording level, resulting in the same playback as before the change. To maintain the integrity of the cinema sound equipment, DTS discs of pre 1999 films should be played only after the subwoofer playback level has been restored to the previous 88 dBC level.

Technicians should follow these steps

Equipment needed: DTS Setup Disc, DS1 or Rev. C

1. Verify the cinema processor is correctly calibrated. Follow the manufacturer’s recommendations for proper playback levels (when using a DTS-6AD, see 4. Below).
2. Using the DTS Setup Disc, adjust the DTS output levels as normal: L, C, R to 85dB SPL and surround(s) to 82 dB SPL* in the theater.
3. Adjust the DTS-6 or DTS-6D subwoofer to 91dBC SPL in the theater.
4. When adjusting the output of the DTS-6AD Cinema Processor using the internal pink noise generator, adjust the DTS subwoofer to 85dBC SPL in the theater.

NOTE
* It is recommended that the subwoofer amplifier be turned off during the surround level adjustments. Turn the subwoofer amp back on after the surround levels have been set.

So for anything released before 1999, the right level for the subwoofer channel is 88 dBC SPF, putting it (in terms of pink noise calibration) 3 dB, rather than 6 dB, louder than L/C/R. This will result in the actual film audio (for anything pre-1999) playing the subwoofer channel at the correct level relative to the other channels.

One thing that's probably worth highlighting here, as it's likely to cause confusion, I think (it's certainly confused the hell out of me):

4. When adjusting the output of the DTS-6AD Cinema Processor using the internal pink noise generator, adjust the DTS subwoofer to 85dBC SPL in the theater.

Here's the calibration section from the manual, which expands slightly upon that discrepancy between internal (generated by the DTS-6AD) and external (played off a disc) pink noise:

5.3.1 SPL ADJUSTMENT

(...)

• Repeat procedure for all channels, set to the following SPLs:
  
  Left, Center, Right 85 dBC
  Left & Right Surrounds: 82 dBC
  Subwoofer: * 85 dBC
  
  * NOTE: The above levels are for use with the internal pink noise generator. When playing back a DTS Set-up or Empirical Disc, the subwoofer level should be 91 dBC. This setting conforms the subwoofer level to the SMPTE RP200 standard of having +10dB in-band gain in relation to the screen channels.
  

(...)

5.3.4. DTS SUBWOOFER LEVEL

(...)

• If a RTA is not available, the DTS subwoofer can be measured using a SPL meter.  The SPLmeter must be set for “C” weighting and “slow.” It should read about 91 dBC when the (DTS Setup Disc) subwoofer pink noise is playing in theater. Take measurements in different areas ofthe auditorium to prevent subwoofer from being adjusted too loud.
  

For some reason I haven't yet worked out for sure, the internal pink noise generator of the DTS-6AD appears to produce pink noise for the subwoofer channel that's quieter than the pink noise that the DTS Setup Disc (or an empirical disc) would generate for that channel. I think I'm going to post separately underneath this to expand upon this a bit; I can't decide if there's actually a good reason for it or if it's just incidental or a necessity of how the DTS-6AD generates pink noise. There does seem to be some key piece of information that I'm not yet finding, because I found reference to the exact same discrepancy between internal and non-internal pink noise measurements in a home theatre context (saying that you should set the subwoofer level to the same as the full-band speakers if using internal pink noise, but if you were using non-internal pink noise, you should set it to +10 dB above them). But in any case, if you don't use the internal pink noise generator on the DTS-6AD, the calibration of the DTS-6AD is exactly the same as the DTS-6D and DTS-6 before it: 91 dBC SPL.

EDIT: I think I might have just confirmed my "I think this is to do with the subwoofer being band limited" hypothesis (detailed in the next post) after all. I finally found some good info on this in a hilariously exasperated forum post: https://www.head-fi.org/threads/testing-...t-14579987

85dBC is the level measured with an SPL meter with the main/screen speaker outputting 20Hz-20kHz pink noise (at -20dBFS). Our LFE is level calibrated to +10dB
in-band gain relative to our main/screen speaker. The "In-band" part means that our LFE/Sub should be outputting 10dB more 20Hz-120Hz than the amount of 20-120Hz our main speaker is outputting. The issue should hopefully now be obvious, our main speaker is not outputting 85dBC of 20Hz-120Hz, it's outputting 85dBC of 20Hz-20kHz. If we remove the 121Hz-20kHz output of our main speaker (so that it's only outputting our required 20Hz-120Hz), it's output level will obviously be lower, it will be approx 81dBSPL and our sub is then calibrated 10dB higher, which is about 91dB. In practice, we wouldn't try and "remove the 121-20kHz", we'd just use an RTA to measure the 20Hz-120Hz portion of the main speaker's output. In other words, your equation should read: "85dBC - (the 121Hz -20kHz band) +10dB = 91dB".

In other words, the reason for the difference is that one way is comparing only the "in-band" level that's available to the subwoofer (most likely 20-120 Hz, I assume, although DTS actually has the sub range end at 80 Hz in practice) and the other way is comparing the band-limited subwoofer output against the full-range (20-20000 Hz) output of the centre channel, a non-equivalence which means you need to boost the sub more to compensate for the lack of higher frequencies in its output. The internal pink noise must be comparing band-limited sub to band-limited centre, or doing it the other way around and sending full band pink noise to both (which wouldn't be possible with pink noise from a disc since it's cutting the subwoofer at 80 Hz), hence both measuring the same at proper calibration, but the DTS Setup Disc and Empirical Disc output full bandwidth L/C/R and band-limited sub so it's not directly comparable. That's more or less what I thought but given some clarity compared to how I tried to explain it! Although the numbers would at face value appear to suggest full band pink noise being sent to both C and sub, I'd be surprised if a cinema sub was expected to play full band properly for measurement purposes (as I mentioned previously), so maybe it instead uses pink noise that's band-limited for all channels but at a higher gain than the setup discs to compensate, meaning that it's now 85 dBC of 20-80 Hz (or possibly 20-120 Hz) on all channels instead of 85 dBC of 20-20000 Hz or whatever that's presumably on the setup discs. Regardless, that explains the discrepancy!

[pipefan413]

Right, as I mentioned, this is the main thing that I can't quite get my head around; there may be some on here who know a lot more than I do about amplification, electrical vs acoustic calibration, and other things that might put them in a better position to understand this, in which case please do gimme a shout and tell me what I'm missing because it drives me nuts not getting something like this!

Basically, the DTS-6AD cinema processor can be calibrated either by using the pink noise off a CD (either the "DTS Setup Disc" or "Empirical Disc") *or* by using pink noise generated internally within the DTS-6AD itself. Depending on which you're using, the SPL (sound pressure level) measurements for the subwoofer channel specifically are different at the point of correct calibration: with internal pink noise it should read 85 dBC, but with external (off a disc) pink noise it should read 91 dBC. And I don't fully understand why.

I'm guessing that it must have *something* to do with where the pink noise lies in the signal processing chain. If it's being played off a disc, then it's subject to all the usual processing characteristics of DTS audio, meaning for instance that it's band-limited to 80 Hz; I wonder if perhaps the internally generated DTS-6AD pink noise instead allows for full-band or maybe 120 Hz limited pink noise instead, which would make it perceptually louder as long as the subwoofer can actually play those frequencies back. I don't imagine full band would be that useful as a control signal since we're talking about sending it to subwoofers, but subwoofers would definitely be capable of playing up to 120 Hz because Dolby Digital and SDDS both use that cutoff instead of the 80 Hz used by DTS. Note: SDDS apparently doesn't actually band-limit its subwoofer channels at all (so they're technically full range, according to the SDDS print master guidelines), but in practice, soundtracks should only contain 120 Hz and below, I think.

EDIT (2): I think I might have just confirmed my "I think this is to do with the subwoofer being band limited" hypothesis after all. I finally found some good info on this in a hilariously exasperated forum post: https://www.head-fi.org/threads/testing-...t-14579987

85dBC is the level measured with an SPL meter with the main/screen speaker outputting 20Hz-20kHz pink noise (at -20dBFS). Our LFE is level calibrated to +10dB
in-band gain relative to our main/screen speaker. The "In-band" part means that our LFE/Sub should be outputting 10dB more 20Hz-120Hz than the amount of 20-120Hz our main speaker is outputting. The issue should hopefully now be obvious, our main speaker is not outputting 85dBC of 20Hz-120Hz, it's outputting 85dBC of 20Hz-20kHz. If we remove the 121Hz-20kHz output of our main speaker (so that it's only outputting our required 20Hz-120Hz), it's output level will obviously be lower, it will be approx 81dBSPL and our sub is then calibrated 10dB higher, which is about 91dB. In practice, we wouldn't try and "remove the 121-20kHz", we'd just use an RTA to measure the 20Hz-120Hz portion of the main speaker's output. In other words, your equation should read: "85dBC - (the 121Hz -20kHz band) +10dB = 91dB".

I didn't really find a definitive answer to what I'm wondering here but I did find some discussion about calibrating a *home* theatre system, which said that using internal pink noise meant you should calibrate a subwoofer to the same level as your full band speakers but if you were using externally sourced pink noise it should be +10 dB higher: https://forums.audioholics.com/forums/th...st-1240815. This is exactly the same instructions given in the DTS-6AD manual: 85 dBC is the same as L/C/R, and that's what they tell you to set it to if using the internal pink noise, but if you're playing pink noise off a disc it says to set it at 91 dBC instead (6 dB louder than L/C/R). So maybe that's all it is: 20-80 Hz pink noise vs full-range (or 20-120 Hz) pink noise?

As a counterpoint to that possibility, I also found multiple references to there just being different standards for test tones depending on the pink noise signal that was being used, and it might simply be that there's no real rhyme or reason to it. See here, for instance: https://forums.audioholics.com/forums/th...ost-167568.


Another thing I can't quite get my head around right now is the instructions in the post production guide about setting the levels for the actual DTS Tower (studio hardware for mixing/mastering cinema DTS). Check this out:

Appendix A

Subject: Setting Sound Pressure Levels

Setting the Sound Pressure Levels (SPL) in theaters and studios is a simple procedure, and should be done routinely. Follow these simple guidelines:

DTS-6 or DTS-6D (Theater) SPL calibration:

Remove the front panel, which covers the output trim potentiometers. Set your SPL meter for C Weighting, Slow Response. Insert the “Rev C” or “DS1” test disc and allow the disc to load. The first thing heard will be 1 kHz across all channels on the “Rev C” test disc only (the “DS1” only has pink noise). Press the small increment button on the lower right twice to change to the pink noise signals on individual channels. The disc will then automatically step through each channel. Adjust the six “6 track” potentiometers for the following readings on your SPL meter:

Left: 85 dB SPL
Left Surround: 82 dB SPL
Center:  85 dB SPL
Right Surround: 82 dB SPL
Right: 85 dB SPL
Sub Woofer: 91 dB SPL

DTS Tower (Studio) SPL calibration:

Using a 1 kHz tone, align the Tower according to the instructions in the respective manual. Once this is done, send pink noise at 0 VU into the Tower. Set your SPL meter for C Weighting, Slow Response. Adjust the “To Monitor” potentiometers for the following readings on your SPL meter:

Left: 85 dB SPL
Left Surround: 82 dB SPL
Center:  85 dB SPL
Right Surround: 82 dB SPL
Right: 85 dB SPL
Sub Woofer: 88 dB SPL

The SMPTE RP200:1999 change meant that the subwoofer level in the actual encoded DTS track would be 3 dB quieter than it was previously, but if they were monitoring it at 88 dB instead of 91 dB then you'd expect this to result in them mixing it 3 dB too *loud* (not quieter) since they'd be hearing it 3 dB quieter than it would end up in theatres. But it might be that there's more to that.

Given that one of the main purposes of the reduction in subwoofer channel level was "to enable the use of one master recording when transferring to all three digital sound processes", I take it that the 6-track format-agnostic master that was actually fed to the DTS Tower had its LFE track 3 dB quieter from 1999 onwards, rather than it simply being that the DTS Tower would be used to create an output subwoofer channel that was 3 dB quieter. Perhaps, if I'm understanding that correctly, it means that before the change they would calibrate their DTS Towers to 85 dBC SPL (same as L/C/R) which would then get boosted by 3 dB on playback, possibly to account for them mixing in a smaller room than a theatre (I dunno though because they monitor the surrounds at -3 dB and those are played back at -3 dB in the theatre, so no difference there). Post-change, the calibration becomes 88 dBC SPL instead, so they're now taking an input that's 3 dB lower than it was before 1999 but monitoring it 3 dB louder than before, meaning that once it sounds balanced on the monitor subwoofer, it'll actually be stored in the output 6-track master 3 dB quieter than a pre-1999 soundtrack, needing to be boosted 6 dB instead of 3 dB overall when played back in the theatre.

I could be way off on that. Certainly, the assertion that this change to the SMPTE standard would "provide playback consistency in theatres" seems to be null and void if they were just ultimately turning it up by 3 dB to offset the 3 dB reduction compared to before the change, surely? I'm confused by the relative levels being measured in the theatre during calibration as well, because according to the manual for the DTS-6AD, the subwoofer channel should be *calibrated* to +10 dB in-band gain higher than the screen channels (L/C/R), which is weird if it's also true that the subwoofer channel in the recorded master was supposedly *created* at +10 dB in-band gain higher than the screen channels. Since calibrating the subwoofer pink noise from a DTS Setup Disc to +10 dB in-band gain above L/C/R is ostensibly achieved by measuring its SPL as 91 dBC, then surely the end result is that the subwoofer channel of any actual recorded DTS soundtrack (as opposed to just pink noise used for calibration) is ultimately played back 6 dB *louder* (+16 dB in-band gain) than the new SMPTE RP200:1999 standard intended, which is exactly the same level it was at before the change? The post production manual even specifies that...

Increasing the DTS subwoofer level compensates for the new lowered recording level, resulting in the same playback as before the change.

This is confusing to me because it seems like it contradicts the post production guide's assertion that...

To comply with this new standard, all DTS films released in North America after January 1, 1999 have been transferred into the DTS digital process with subwoofer at 10dB in-band gain. The SMPTE RP200 logo is clearly visible on the discs of these films.

Surely if it's "transferred into the DTS digital process" at +10 dB, but then calibrated up another +6 dB for playback, then DTS have basically just come up with a simple workaround to avoid the "provide playback consistency in theaters" goal of SMPTE RP200:1999. Like, yes, this technically does mean that after calibration to 91 dBC SPL the in-band gain of the subwoofer channel is 10 dB above the screen channels if measured purely on the basis of pink noise used for calibration, but in practical terms, they just mastered it 3 dB lower so that it'd play back at the exact same level as before when calibrated to 88 dBC SPL. In other words, SMPTE yelled, "Will you turn that s**t down?!" and DTS yelled back, "No! You're not my real dad!"

Unless, of course, I'm completely misunderstanding some fundamental part of why the DTS Tower calibration was 88 dBC SPL from 1999 onwards. Basically, this could easily be answered if I could see a pre-1999 DTS post production guide, but the only one I've been able to find so far is the 1999 one hosted on film-tech.

The thing is, the SDDS guidelines appear to be saying the exact same as the DTS calibration instructions: to turn the subwoofer channels up to 91 dBC vs the 85 dBC of the main screen channels. So this doesn't appear to be unique to DTS, which would support the idea of standardisation across all three digital formats. I haven't yet dug out Dolby Digital (cinema version) documentation, which might prove useful, but there was a fair bit of information buried in the "subwoofers" section of the SDDS guidelines that had me wondering if I was just on the fringes of finally getting the penny to drop except that I'm not fully understanding what I'm reading there. There's a lot of stuff distinguishing between electrical and acoustic reference levels, which seems like it might tie in with the DTS documentation repeatedly insisting that you should calibrate and set levels using the internal pink noise and a real time analyser (RTA) instead of a test disc and SPL meter... but yeah, I really don't know.

I'm quite sure that this is because I don't have an education in audio engineering and if I did it would be plainly obvious why there's a difference between measuring internally generated pink noise and external, decoded pink noise, but I've been faffing about with this for too long already and I think I need to admit defeat and yell for an adult.

Any takers?

EDIT: Just found a tech note referenced in the SDDS manual, so going to have a read through that in case it contains any info not in the main guidelines document. It's here, for anybody interested: http://www.film-tech.com/warehouse/manua...051701.pdf

Almost certainly relevant:

3. Setting the subwoofer acoustical gain using a screen speaker as a reference.
(...)
To make this adjustment properly, a multi-channel real time analyzer must be used. Analyzer bands in the flat-response region of a screen speaker (generally, the center speaker) are taken as a reference. These bands will not individually measure 85dB SPL, but will be somewhat less, depending on the bandwidth of the analyzer’s bands, typically about 70 dB SPL for a 1/3-octave analyzer as required by SMPTE 202M. This flat-response region, between the low frequency roll off caused by the loudspeaker cabinet and the high frequency roll off due to the screen and X-curve, is referred to as the in-band (acoustical) response of the monitor speaker. The subwoofer electrical gain is adjusted such that the analyzer channels in it’s in-band region are 10dB greater than those in the screen speaker’s in-band region. This is referred to as “10dB of in-band gain” (subwoofer level relative to each screen speaker level).



The measurement is made acoustically, not electrically. Typically, each band of the analyzer in the pass band of the subwoofer will then measure about 80 dB SPL.

So this seems to align very well, because if the full range (e.g. centre) channels are running a given band at 70 dB, and the sub is running 80 dB, that certainly sounds like "+10 dB in-band gain" to me. And then it goes on to say...

If a screen speaker and the LFE subwoofer loudspeaker are each measured with a wide band SPL meter, the subwoofer will typically measure approximately 5.5 dB higher.

... which would probably be why the discrepancy between 85 dBC for internal pink noise and 91 dBC for on-disc pink noise (6 dB difference, very very close to 5.5 dB as stated here).

I feel like I'm just about there but not quite getting to some golden nugget of truth that would let me reach complete understanding. Bloody frustrating.

[zoidberg]

I was working with little-endian on this last year but my research got derailed, if I remember there is some mystery wrt how DTS handles the 80Hz crossover on the surrounds. But yes there are 2 different 'flavours' of the cinema DTS, pre and post RP200 with their different subwoofer levels. Baby boom/LFE channels, going back to 70mm mag have always been boosted during playback compared to the main full-bandwidth speakers in order to maximise the potential dynamic range (peak main channel 105dB, peak LFE 115dB, max overall 120dB lol). Whereas all the other systems used 10dB of in-band gain, DTS was an outlier in that it used less playback gain initially (perhaps to maximise the SNR, maybe just because they could) but was brought into line by the new SMPTE standard. So they knocked 3dB off the recorded level and added it back during playback.

The 10dB in-band gain refers to the fact that when correctly calibrated the subwoofer's playback level, for it's operating frequency range will be 10dB higher than the equivalent level for each of the individual 3 front speakers, when measured with an RTA.

Concerning the test levels I think it's purely a case of different methods achieving the same goal. Both methods will result in correct playback levels but running the test disc will confirm 'in use' levels are correct.

[pipefan413]

I was working with little-endian on this last year but my research got derailed, if I remember there is some mystery wrt how DTS handles the 80Hz crossover on the surrounds. But yes there are 2 different 'flavours' of the cinema DTS, pre and post RP200 with their different subwoofer levels. Baby boom/LFE channels, going back to 70mm mag have always been boosted during playback compared to the main full-bandwidth speakers in order to maximise the potential dynamic range (peak main channel 105dB, peak LFE 115dB, max overall 120dB lol).

Holy crap! Aye I saw something about this going back to Todd-AO in the SDDS Tech Note:

Why we do it this way.

The reason for lowering the electrically recorded level of the subwoofer channel and making it up by turning up the playback gain dates back to 70mm film. The subwoofer recording level was lowered to prevent saturation of the magnetic track and the playback gain was increased to compensate. The loss of signal to noise performance was inconspicuous because the subwoofer signal was sent through a low pass filter and did not reproduce hiss. This level difference convention has been retained in the digital world, where it serves to give additional effective headroom for the playback of low frequency sound effects through the subwoofer loudspeaker.

I think I misread part of the manual in a small but significant way, which would go some way to explaining my confusion over this. Looky here:

5.3. FINAL SOUND LEVEL ADJUSTMENTS

5.3.1. SPL ADJUSTMENT
(...)
- Repeat procedure for all channels, set to the following SPLs:
Left, Center, Right: 85 dBC
Left & Right Surrounds: 82 dBC
Subwoofer: *85 dBC

* NOTE:
The above levels are for use with the internal pink noise generator. When playing back a DTS Set-up or Empirical Disc, the subwoofer level should be 91 dBC. This setting conforms the subwoofer level to the SMPTE RP200 standard of having +10dB in-band gain in relation to the screen channels.

(...)

5.3.2. CHECKING DTS DIGITAL LEVELS

Insert DTS  Setup  Disc. When the disc is read, the DTS-6AD should automatically switch  to DIGITAL format. Each channel is announced before pink noise is heard in theater & boothmonitor. Press F4 to advance to the next program. The Setup Disc is used to verify levels only. Levels are set using the internal pink noise generator.

Verify digital levels:
Left, Center, Right: 85 dBC
Left & Right Surrounds: 82 dBC

(digital subwoofer next page)

(...)

5.3.4. DTS SUBWOOFER LEVEL

- The most accurate way to measure DTS digital subwoofer level is to use a RTA.

- The DTS Setup disc must be used, it supplies digital subwoofer pink noise to the auditorium.

- The DTS digital subwoofer should have +10dB in-band gain as compared to the screen channels. The graph below shows the pattern that should be exhibited by the RTA when the (DTS Setup Disc) subwoofer pink noise is playing in the theater.



- If a RTA is not available, the DTS subwoofer can be measured using a SPL meter. The SPL meter must be set for “C” weighting and “slow.”  It should read about 91 dBC when the (DTS Setup Disc) subwoofer pink noise is playing in theater. Take measurements in different areas ofthe auditorium to prevent subwoofer from being adjusted too loud.

Main difference from what I said before: almost the whole way through, the DTS-6AD manual tells you explicitly to use the pink noise off the setup disc for the subwoofer, NOT the internal pink noise. There is one exception where it is NOT consistent with this: *before* it repeatedly says to use the disc for the subwoofer, it says (under the "SPL ADJUSTMENT" section) that the level should be 85 dBC for the subwoofer *if using the internal pink noise* which is what lost me. I... don't know why it appears to be saying, "don't use the internal pink noise for the subwoofer" whilst also saying "here's what it should measure if you use the internal pink noise for the subwoofer". What it might be is that the order these statements appear in is (and I'll paraphrase here intentionally so that I'm wording it more clearly than it actually does in the manual):

1. (From "SPL ADJUSTMENT") When setting levels, use internal pink noise generator and measure with an SPL meter (set to C weighted, slow). Subwoofer should be set to 85 dBC SPL when measuring in this way.

2. NOTE: Later, when you use the DTS Setup Disc and an RTA to verify the levels you've set, the pink noise it plays should measure 91 dBC SPL - NOT 85 dBC like the internal pink noise does.

3. (From "DTS SUBWOOFER LEVEL") To check that the levels you set earlier with the internal pink noise and SPL meter are actually correct for the subwoofer, switch to using an RTA instead of an SPL meter, and verify the levels with pink noise from a disc (like the DTS Setup Disc or Empirical Disc). The RTA should show that the subwoofer is 10 dB louder in its ~20-80 Hz band compared to the centre channel. If you don't have an RTA, the SPL measurement for this pink noise from the disc should measure 91 dBC from your SPL meter.

In other words, I think they're just using the internal pink noise generator and SPL meter to get it roughly right first (since this will be less accurate and measures it in isolation rather than specifically measuring it relatively against L/C/R), and then after that to make sure it's actually meeting the specific +10 dB in-band requirement of SMPTE RP200:1999, they're saying to measure it using the pink noise from a disc (which will actually be encoded as DTS, I would think, and therefore be decoded according to the standards the DTS-6AD applies, which seems to include boosting the sub by 6 dB... I'll come back to this in a moment).


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. This is more or less how the -3 dB surround attenuation and +6 dB subwoofer boost conclusions were reached. For pre-1999 films, the DTS-6AD therefore requires -3 dB attenuation on the subwoofer to bring it back to pre-1999 calibration, whereas the foobar2000 plugin needs a +3 dB boost to arrive at the same level.

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. Perhaps it doesn't apply that boost when outputting internally generated pink noise, because it probably isn't actually decoding DTS when it outputs pink noise, but simply generating and throwing out a signal at a uniform level. Initially I thought it might be something to do with the in-band vs full-band thing, but that probably wouldn't make sense because if it was throwing the same amount of pink noise at full-band range that the disc is throwing at subwoofer-band range (20-80 Hz) then it'd be much louder, not quieter. Alternatively, it might simply be a safety measure: maybe they were worried about people blowing speakers if they incorrectly assumed that the DTS-6AD output still needed the +6 dB boost applied to the subwoofer channel, so they lowered the gain on the internal pink noise to remove that so you'd not burst your speakers even if you accidentally calibrated it 6 dB too loud. But I dunno about that because they then go on to tell you to check it with the DTS Setup Disc, at which point - if you *had* calibrated wrong - it'd be playing pink noise at 97 dBC instead of 91 dBC. So I don't think it's necessarily deliberate, and probably more likely something more like what I said a moment ago about it being because the pink noise isn't going through the same processing as the pink noise off an actual DTS-encoded disc.

I'd be curious to actually take measurements of the pink noise directly from the DTS-6AD, but I have thus far resisted the urge to buy one myself. Regardless, it would be unlikely to affect the conclusions drawn thus far: it does seem to be the case that pre-1999 films need subwoofer boosted by 3 dB and 1999+ films need it boosted by 6 dB when decoding in the foobar2000 APTX-100 plugin, and all films regardless of release date should have the surrounds attenuated by 3 dB. And I can't afford it right now anyway.

Whereas all the other systems used 10dB of in-band gain, DTS was an outlier in that it used less playback gain initially (perhaps to maximise the SNR, maybe just because they could) but was brought into line by the new SMPTE standard. So they knocked 3dB off the recorded level and added it back during playback.

The 10dB in-band gain refers to the fact that when correctly calibrated the subwoofer's playback level, for it's operating frequency range will be 10dB higher than the equivalent level for each of the individual 3 front speakers, when measured with an RTA.

It's just weird to me that they describe it as two different and seemingly contradictory things, at once saying that it's +10 dB in-band at *playback* and also that it's "transferred into the DTS digital process" at +10 dBC in-band gain. If it actually is boosted during playback then surely both statements cannot possibly be true at the same time.

If I can momentarily ignore that every other piece of info suggests that it is indeed boosted during playback, now that I actually understand what "in-band" means (comparing only the relevant frequency band rather than limited vs full range), I suppose it's possible that it *is* +10 dB both at mixing/mastering stage and at playback stage; it could be "transferred into the DTS digital process" at +10 dB in-band (20-80 Hz) compared to C, stored in the codec at that level, and played back at that same level in the end. But that contradicts all this other info which indicates that it is actually boosted at playback, if calibrated correctly, following the same basic methodology as legacy analogue 70 mm magnetic 6-track (Todd-AO). For it to be true that it's transferred in and played back at +10 dB in-band *and* that it's boosted during playback, it seems to me that it must therefore be *stored* at a lower level than this +10 dB standard when encoded into the codec. In other words, they digitise, mix and master it at +10 dB in-band compared to the centre channel, encode it quieter than that (by the looks of it, at -6 dB), but then boost it back to +10 dB in-band again during playback. If that's what it is, then no bloody wonder I was confused by all this.

Phew.

Concerning the test levels I think it's purely a case of different methods achieving the same goal. Both methods will result in correct playback levels but running the test disc will confirm 'in use' levels are correct.

Aye, I guess it's just more precise using RTA hence they keep insisting you *set* calibration using RTA and *check* it using setup disc. I just wish the manual was clearer about this, because it reads like it's telling you one thing then immediately telling you another thing about how to correctly set and measure the sub. One moment it feels like "use the internal pink noise" then after that "don't use the internal pink noise, use the disc", but I think I now get what it's actually meaning. The phrasing in the instructions is a bit of a mess imo.


Apart from me wondering about this pink noise level discrepancy, I think the main thing I'm still curious about is exactly why you'd actually monitor the sub channel 3 dB quieter off the DTS Tower. Maybe it is just literally mixing/mastering in a smaller room so they would perceive it as louder compared to large cinema room and cut 3 dBC to account for the perceptual difference.

Perhaps at some point I'll try to get my hands on a DTS-6AD and DTS Setup Disc and have a play around, but that feels like an expensive experiment!

[zoidberg]

They're probably using the old subwoofer value of 88dB for mixing/monitoring, knowing that the encode stage will adjust the level for playback to RP200 spec.

[pipefan413]

They're probably using the old subwoofer value of 88dB for mixing/monitoring, knowing that the encode stage will adjust the level for playback to RP200 spec.

I considered that but I thought it strange that you'd not just monitor it at the level you intend it to be heard at. Surely it'd be odd to have to constantly think "this is quieter than it will actually sound in the end"? But I dunno, it might just be exactly that.

[zoidberg]

But they are listening to it at the correct level

[pipefan413]

But they are listening to it at the correct level

*Goes for a lie down*

[zoidberg]

As far as I can tell the subwoofer level adjustment to meet SMPTE RP200 is applied during the creation of the CD-ROMs. It isn't applied at the studio stage, which behaves as pre-RP200 so 88dB subwoofer for 10db in-band gain over the main channels. Once encoded to disc, theatrical playback requires the subwoofer level to be raised to maintain the in-band gain so 91dB or thereabouts.
I'm curious as to where you're going with this, none of this is really beneficial to us converting cinema DTS for our projects. The main thing is getting the levels off the discs correct

[pipefan413]

As far as I can tell the subwoofer level adjustment to meet SMPTE RP200 is applied during the creation of the CD-ROMs. It isn't applied at the studio stage, which behaves as pre-RP200 so 88dB subwoofer for 10db in-band gain over the main channels. Once encoded to disc, theatrical playback requires the subwoofer level to be raised to maintain the in-band gain so 91dB or thereabouts.

Which would make sense to me and would have been exactly what I'd assume had I not seen some seemingly contradictory stuff in the documentation. Again, the only reason I thought otherwise at any point was that the post production guide specifies that they changed it to do two things: standardise it in theatres and standardise it at master so you could transfer the same 6-track into Dolby Digital, DTS, and SDDS without manipulating levels or whatever. They specifically say that it has +10 dB in-band when it's "transferred into the DTS digital process", which doesn't sound to me like it means "after they've transferred it into the DTS process and finished mixing and mastering on a DTS Tower". But maybe it does and that's just a slightly misleading choice of words that I'm chasing away down the garden path.

I'm curious as to where you're going with this, none of this is really beneficial to us converting cinema DTS for our projects. The main thing is getting the levels off the discs correct

Yeah, I know, that's what matters in a practical sense. That's why I put that info up front, it's right there at the start of the top post. I'm just curious about anything beyond that, hence splitting the rest into a separate post etc.

My motivation for understanding is often understanding in and of itself, nothing more. Not everything needs to be for practical purposes, though I do find that often understanding things better does lead to more sophisticated choices being made in hitherto unexpected ways. I appreciate your help, as always.