Poll: Should Sony use DTS-HD Master Audio or Dolby TrueHD?

[Sonny]

Originally Posted by deado

I use eac3to to convert lossless tracks to FLAC (Free Lossless Audio Codec) so I can get full bit-depth and sample rate on my PC (because software players downsample to 48/16).

eac3to tells you everything about the streams on the disc, most specifically the audio. It tells you what it does in the log during the conversion. Here is the relevant parts of the log for The House Bunny: (just an example):

Streams on the disc:



Then during conversion:



What this means is that eac3to is only detecting 16-bits of information in the TrueHD container that contain any info... the rest of the container contains nothing. The log says the same thing for The Dark Knight, and every recent Warner title I've tried. The log also says the same thing for every other Sony title I've tried recently too (Seven Pounds, You Don't Mess with the Zohan, Quarantine), and many more.

I shouldn't have said *every* Sony track as I can't confirm that, but it appears to me as though a majority are only 16-bit. If they were old films that's to be expected, however these examples are not old films.


Real 24-bit tracks, in comparison, say this during conversion:



From my testing, only Sony and Warner are putting 16-bit tracks in 24-bit containers, other studios which have 16-bit tracks put them in 16-bit containers.

And when I play back the .mkv files with the FLAC audio, ReClock (which displays exactly what is being output on my sound card via WASAPI Exclusive mode), says 48/16 for the Warner tracks, and 48/24 for the real 24-bit tracks.

Nice work.

[Blu Titan]

Originally Posted by RebelJeeper

I like the DTS option personally...seems to be more impressive to me. But what do I know?

You dollar is as worthy as the expert's.

[bluseminole]

Originally Posted by deado

I use eac3to to convert lossless tracks to FLAC (Free Lossless Audio Codec) so I can get full bit-depth and sample rate on my PC (because software players downsample to 48/16).

eac3to tells you everything about the streams on the disc, most specifically the audio. It tells you what it does in the log during the conversion. Here is the relevant parts of the log for The House Bunny: (just an example):

Streams on the disc:



Then during conversion:



What this means is that eac3to is only detecting 16-bits of information in the TrueHD container that contain any info... the rest of the container contains nothing. The log says the same thing for The Dark Knight, and every recent Warner title I've tried. The log also says the same thing for every other Sony title I've tried recently too (Seven Pounds, You Don't Mess with the Zohan, Quarantine), and many more.

I shouldn't have said *every* Sony track as I can't confirm that, but it appears to me as though a majority are only 16-bit. If they were old films that's to be expected, however these examples are not old films.


Real 24-bit tracks, in comparison, say this during conversion:



From my testing, only Sony and Warner are putting 16-bit tracks in 24-bit containers, other studios which have 16-bit tracks put them in 16-bit containers.

And when I play back the .mkv files with the FLAC audio, ReClock (which displays exactly what is being output on my sound card via WASAPI Exclusive mode), says 48/16 for the Warner tracks, and 48/24 for the real 24-bit tracks.

Seems legit to me!

[ClaytonMG]

Originally Posted by deado

I use eac3to to convert lossless tracks to FLAC (Free Lossless Audio Codec) so I can get full bit-depth and sample rate on my PC (because software players downsample to 48/16).

eac3to tells you everything about the streams on the disc, most specifically the audio. It tells you what it does in the log during the conversion. Here is the relevant parts of the log for The House Bunny: (just an example):

Streams on the disc:



Then during conversion:



What this means is that eac3to is only detecting 16-bits of information in the TrueHD container that contain any info... the rest of the container contains nothing. The log says the same thing for The Dark Knight, and every recent Warner title I've tried. The log also says the same thing for every other Sony title I've tried recently too (Seven Pounds, You Don't Mess with the Zohan, Quarantine), and many more.

I shouldn't have said *every* Sony track as I can't confirm that, but it appears to me as though a majority are only 16-bit. If they were old films that's to be expected, however these examples are not old films.


Real 24-bit tracks, in comparison, say this during conversion:



From my testing, only Sony and Warner are putting 16-bit tracks in 24-bit containers, other studios which have 16-bit tracks put them in 16-bit containers.

And when I play back the .mkv files with the FLAC audio, ReClock (which displays exactly what is being output on my sound card via WASAPI Exclusive mode), says 48/16 for the Warner tracks, and 48/24 for the real 24-bit tracks.

That actually kind of sucks. I know at one time Sony was doing 24-Bit tracks and just kind of assumed they had kept with that (why change?). But if they changed to DTS-HD MA, would they really change their bit depth?

[saprano]

Originally Posted by deado

I use eac3to to convert lossless tracks to FLAC (Free Lossless Audio Codec) so I can get full bit-depth and sample rate on my PC (because software players downsample to 48/16).

eac3to tells you everything about the streams on the disc, most specifically the audio. It tells you what it does in the log during the conversion. Here is the relevant parts of the log for The House Bunny: (just an example):

Streams on the disc:



Then during conversion:



What this means is that eac3to is only detecting 16-bits of information in the TrueHD container that contain any info... the rest of the container contains nothing. The log says the same thing for The Dark Knight, and every recent Warner title I've tried. The log also says the same thing for every other Sony title I've tried recently too (Seven Pounds, You Don't Mess with the Zohan, Quarantine), and many more.

I shouldn't have said *every* Sony track as I can't confirm that, but it appears to me as though a majority are only 16-bit. If they were old films that's to be expected, however these examples are not old films.


Real 24-bit tracks, in comparison, say this during conversion:



From my testing, only Sony and Warner are putting 16-bit tracks in 24-bit containers, other studios which have 16-bit tracks put them in 16-bit containers.

And when I play back the .mkv files with the FLAC audio, ReClock (which displays exactly what is being output on my sound card via WASAPI Exclusive mode), says 48/16 for the Warner tracks, and 48/24 for the real 24-bit tracks.

Thanks for that info. oh well. i guess sonys going on the half assed list with warner.

[PeterTHX]

The reason often they go with 16-bit is because they want to offer things like the French, Portugese and other tracks in lossless as well so they can create one master for multiple markets. 24-bit takes up pretty much double the space, not to mention more bandwidth and the goal is to not impact the video as much.

[saprano]

Originally Posted by PeterTHX

The reason often they go with 16-bit is because they want to offer things like the French, Portugese and other tracks in lossless as well so they can create one master for multiple markets. 24-bit takes up pretty much double the space, not to mention more bandwidth and the goal is to not impact the video as much.

Yes but as deado pointed out they put it in 24bit containers. why not just do a true 16bit track?

By the way im not going to even pretend like i know how this bit stuff works. im just going off of what deado posted.

[deado]

Originally Posted by PeterTHX

The reason often they go with 16-bit is because they want to offer things like the French, Portugese and other tracks in lossless as well so they can create one master for multiple markets. 24-bit takes up pretty much double the space, not to mention more bandwidth and the goal is to not impact the video as much.

Yes that is correct, most new Sony releases have TrueHD in multiple languages.

[PeterTHX]

Originally Posted by deado

Yes that is correct, most new Sony releases have TrueHD in multiple languages.

From my testing, only Sony and Warner are putting 16-bit tracks in 24-bit containers, other studios which have 16-bit tracks put them in 16-bit containers.

Well, yes, I know that Warner is 16-bit. I have yet to see a 24-bit "containers" for Warner though.

Easy way for me to tell is that Warner titles and some Sony run around 1.7 Mbps with 3.3Mbps peaks. 24-bit tracks are 2.7Mbps with 5.5Mbps peaks, for instance the majority of Paramount titles.

[deado]

Warner definitely has 24-bit containers, because eac3to says "Reducing depth from 24 to 16" (does a second pass because it only detects 16 bits in the container).

I even looked at the log for Troy HD-DVD (the first release), and it says the same thing.

[CAB]

And I learn even more stuff.

Makes me further wonder what is being used as masters if the consumer is keying off the stats on the encode instead of what is in the package. I'm also reminded of the issues created when bit depth differs between releases from different regions. Is it really 24-bit? If it is 24-bit in the package, is it just an upsampled version of the same 16-bit master used elsewhere?

Foggy!

[Kris Deering]

Paidgeek had chimed in on bit depth awhile back. He said that they were going to most likely settle on 20 bit encodes for TrueHD as they didn't feel the LSBs warranted the hit they would take on the video side by including them. Sounds like they've settled on 16 bit though instead.

[lgans316]

Originally Posted by Kris Deering

Paidgeek had chimed in on bit depth awhile back. He said that they were going to most likely settle on 20 bit encodes for TrueHD as they didn't feel the LSBs warranted the hit they would take on the video side by including them. Sounds like they've settled on 16 bit though instead.

16-bit and 448 Kbps on many including Casino Royale: CE . Thank God, the rights went to FOX for QoS.

[Cinema Squid]

Originally Posted by saprano

Yes but as deado pointed out they put it in 24bit containers. why not just do a true 16bit track?

As far as I know (and I may be wrong on some of these points - corrections are welcome!), there is no such thing as a "16-bit" TrueHD container - at least as implemented for Blu-ray. TrueHD audio headers do not contain sample bit depth information, so all samples decode to 24-bit values and samples are zero-padded in the least significant bits during decoding if the "real" bit depth is lower.

The only reliable way to determine the bit depth of a TrueHD track is to examine each sample as deado has done with eac3to to see what the minimum number of non-zero bits to represent all the samples would be. However, you can usually make reasonably informed guesses from average/peak raw bitrates to estimate 16-bit vs. something higher like 20- or 24-bit.

Things can get even more complicated, however, since you may occasionally see different effective bit depths allocated across the channels - for example, 24-bits for the fronts and 16-bits for the surrounds. The effective bit depth does not have to remain constant either, so it can vary up and down throughout the film with more or less sample LSB padding occurring in different segments.

Personally, I see this as an space-saving advantage of the TrueHD codec on Blu since, perceptually speaking, there are rapidly diminishing returns on anything above 16-bit for most of the population but the encoders can still crank up the fidelity for more challenging and dynamic segments just to be on the safe side. Those rare, blessed individuals with both golden ears and platinum equipment may disagree...

[srrndhound]

Originally Posted by Cinema Squid

As far as I know (and I may be wrong on some of these points - corrections are welcome!), there is no such thing as a "16-bit" TrueHD container - at least as implemented for Blu-ray. TrueHD audio headers do not contain sample bit depth information, so all samples decode to 24-bit values and samples are zero-padded in the least significant bits during decoding if the "real" bit depth is lower.

The only reliable way to determine the bit depth of a TrueHD track is to examine each sample as deado has done with eac3to to see what the minimum number of non-zero bits to represent all the samples would be. However, you can usually make reasonably informed guesses from average/peak raw bitrates to estimate 16-bit vs. something higher like 20- or 24-bit.

Things can get even more complicated, however, since you may occasionally see different effective bit depths allocated across the channels - for example, 24-bits for the fronts and 16-bits for the surrounds. The effective bit depth does not have to remain constant either, so it can vary up and down throughout the film with more or less sample LSB padding occurring in different segments.

Personally, I see this as an space-saving advantage of the TrueHD codec on Blu since, perceptually speaking, there are rapidly diminishing returns on anything above 16-bit for most of the population but the encoders can still crank up the fidelity for more challenging and dynamic segments just to be on the safe side. Those rare, blessed individuals with both golden ears and platinum equipment may disagree...

Correct on all points.

I would add that the 24-bit output structure makes it easier for post-processing DSPs and DACs to deal with a consistent bit depth, which can vary not only between 16, 20, and 24, but 17, 18, 19...23 --something PCM cannot do normally. Dolby strongly recommended 20 bits as the ideal resolution--no sonic compromise, and saves 1 Mb/s data rate or 1 GB in a typical movie. That's the same space needed to go from 5.1 to 7.1 channels.

[srrndhound]

Originally Posted by Sir Terrence

Actually the original DTS release of Jurassic Park did not have a poor bass response at all. It was done much like the Transformers release. Instead of directing most of the bass to the LFE where it would be boosted 10db, the bass was partially re-directed to the mains. For those using a sub-sat system (all bass directed to the sub) the bass did not get the 10db boost it would have received if it was all directed to the LFE. It sounded weak on those types of systems. Those of us who use full range speakers in the frontal hemisphere, we heard all the bass as it was supposed to be. The LFE was weak, but the bass was all there.

Bass management does not care whether the bass is carried in the LFE or the mains—the results will be the same. If the bass levels change dramatically between full range and sat/sub speakers, then there’s something out of whack in the playback system.

DTS never cooked anything, that is a false rumor and a myth. As far as I know, Dolby recommended the surrounds be reduced by 3db for home video releases. That was not something DTS knew about, and when they did find out, they reduced the volume by the necessary 3db. No cooking here. All mixes that DTS did were approved by the producer or director, so there was no cooking here as well.

Yes, dts has plausible deniability—forgetting to apply the –3dB conversion for home video releases. They laid the blame on the post-production house.

It is true that Dolby recommended that theatrical masters have their surrounds attenuated by 3dB when re-mastered for home. It was decided that this carryover in cinema calibration from the days of mono surrounds could be avoided, and give the home 5.1 platform identical signal vs. SPL across all the main channels, rather than following the cinema practice of requiring both Ls and Rs to be energized (or drive one side 3 dB hotter) to reach the same SPL as any of the L/C/R. Dolby’s involvement in content mastering, theatrical playback systems, and later in home playback format standards allowed a complete view of the calibration landscape, and it was therefore easy to manage the transition to discrete 5.1 into the home and ensure proper calibration with a simple “Surrounds -3dB” button on the DD encoders.

When looking for smoking guns, it’s not easy—these things happen behind closed doors. Dolby noticed these and other audible differences (like EQ) between the masters and the DTS LDs, and wondered why the studios allowed them to sweeten (or otherwise alter) the mixes. If they could do it, Dolby might like to have the same option. So studio folks, Disney among them, were invited to Dolby’s Burbank office to hear these soundtrack differences, and were quite astonished that they existed. “We thought they just encoded what we sent them” was the typical comment. It was about that time the studios began bringing DTS encoding in-house.

Ok, let’s say they forgot—or someone else made an error they didn’t detect. How does that explain the elevated surrounds on music DVDs? For example, Santana Supernatural Live has both DD and DTS tracks. No, they didn’t actually boost the surrounds, the L/C/R are all attenuated 3 dB. That means the surrounds and LFE both sound stronger compared with the DD track. IIRC the same happened on other DVDs by Diana Krall, Faith Hill, Josh Groban. This is years after the LD incident. What button does that?

I remember hearing the story from Universal Music, who had just made their first DVD-A discs, and were naturally very proud of the technical achievement. To show the sonic merits, they played it against an earlier release of the same content on DTS CD. The difference was obvious--the dts one sounded better! No, I didn't hear this first hand.

Ok, so what about the Dave Grusin West Side Story DVD? It was mixed as DD 5.0, no LFE. Plenty of bass. Some time later, it was issued again as a DTS CD. Now all of a sudden it has an LFE track. More bass. Why? How? I sat across from Phil Ramone, who produced it, at a dinner. I had to ask if he wanted it added. He said emphatically, no. And when he heard the check disc, he complained it had too much bass and rejected it--over and over. He wanted 5.0, and made 5.0. Whose finger accidentally hit the “make an LFE track” button? DTS said it was another producer, not DTS.

Originally Posted by Sir Terrence

I guess what DTS is doing is probably no worse than Dolby mentioning that Dolby TrueHD can do 14 channels or more, when the format that it is referenced to can only handle 8.

Both DTS and Dolby use numbers to impress, not just DTS. A little balance might be in order here.

When Dolby states TrueHD can support 14 channels, that is a fact of the codec as based on the 18 Mbps bitrate. It can handle more than that given higher data capacity. The only thing restricting TrueHD to 8 channels is the BD spec, which came after TrueHD was defined. And unlike the way MLP was designed into DVD-A, with a 6-ch cap that could not be altered, BD was designed such that if, someday, the BDA chooses to revise that figure to more than 8, the new discs will play compatibly with existing players and decoders. No hardware will be made obsolete. I think that's more than impressive--it's futureproof. (Yes, to hear the extra channels discretely, a new decoder would be needed.)

[deado]

Originally Posted by Cinema Squid

As far as I know (and I may be wrong on some of these points - corrections are welcome!), there is no such thing as a "16-bit" TrueHD container - at least as implemented for Blu-ray. TrueHD audio headers do not contain sample bit depth information, so all samples decode to 24-bit values and samples are zero-padded in the least significant bits during decoding if the "real" bit depth is lower.

The only reliable way to determine the bit depth of a TrueHD track is to examine each sample as deado has done with eac3to to see what the minimum number of non-zero bits to represent all the samples would be. However, you can usually make reasonably informed guesses from average/peak raw bitrates to estimate 16-bit vs. something higher like 20- or 24-bit.

Things can get even more complicated, however, since you may occasionally see different effective bit depths allocated across the channels - for example, 24-bits for the fronts and 16-bits for the surrounds. The effective bit depth does not have to remain constant either, so it can vary up and down throughout the film with more or less sample LSB padding occurring in different segments.

Personally, I see this as an space-saving advantage of the TrueHD codec on Blu since, perceptually speaking, there are rapidly diminishing returns on anything above 16-bit for most of the population but the encoders can still crank up the fidelity for more challenging and dynamic segments just to be on the safe side. Those rare, blessed individuals with both golden ears and platinum equipment may disagree...

Actually yes I think you are correct - I have seen dozens of DTS-HD MA tracks in 16-bit containers, but come to think of it I don't think I've seen any TrueHD tracks which are 16-bit containers, which would back up what you're saying.

Something else which is interesting - is sometimes eac3to says "Bit depth is approximately (or close to) 16-bit, extracting 16-bits". I think this has occured with some older films released by Fox which are in 24-bit DTS-HD containers but the film doesn't have enough audio info to reach much above 16-bit.

From memory, I *think* Ghosts of Mars said this as well (Sony release, 2001 movie).

Maybe this has to do with the different channels having different bit depth, which complicates things further.

[SlaughterX]

Since more stand alone players seem to be able to convert DDTHD to PCM but not DTS-HD (like the Samsung BD-P1500 I own), I think it would be a better choice to stick with True HD.

[lgans316]

Originally Posted by deado

Actually yes I think you are correct - I have seen dozens of DTS-HD MA tracks in 16-bit containers, but come to think of it I don't think I've seen any TrueHD tracks which are 16-bit containers, which would back up what you're saying.

Something else which is interesting - is sometimes eac3to says "Bit depth is approximately (or close to) 16-bit, extracting 16-bits". I think this has occured with some older films released by Fox which are in 24-bit DTS-HD containers but the film doesn't have enough audio info to reach much above 16-bit.

From memory, I *think* Ghosts of Mars said this as well (Sony release, 2001 movie).

Maybe this has to do with the different channels having different bit depth, which complicates things further.

It's Cinema Squid the author of the widely used BDInfo utility. So it's hard to go wrong.

[deado]

oh, lol. I'll shut up now