Dolby TrueHD vs. DTS-HD Master Audio on HD DVD/Blu-Ray (1 Viewer)

Although we haven't heard any DTS-HD Master Audio lossless discs yet, I'd like to get everyone's opinion on which codec they prefer in terms of theory.

Here is how I understand it. First, Dolby TrueHD (being based on MLP) is not backwards compatible with Dolby Digital Plus/Dolby Digital, meaning that if you playback a TrueHD signal on a device without TrueHD decoding you will get no sound even if that device has standard DD decoding. DTS-HD Master Audio is backwards compatible with DTS, meaning that if you playback a DTS-HD Master Audio soundtrack on a device without DTS-HD Master Audio decoding, you will get standard DTS sound so long as the device can decode standard DTS - but at what cost?

Let's say "X" is the Master.

Dolby TrueHD encodes "X" into a lossless signal which is decoded by a TrueHD decoder during playback. Fairly straightforward.

DTS-HD Master Audio encodes "X" into a lossy signal "Y" and stores the information normally thrown away during lossy encoding as "Z". When played back on a standard DTS decoder, all you hear is "Y", and "Z" is ignored. When played back on a DTS-HD Master Audio decoder, "Z" is added to "Y", theoretically giving you the original signal "X" (hence lossless). One big problem I see with this. Since Standard DTS is a lossy signal, it is likely some artifacts will be present in the lossy "Y," however those artifacts obviously can't be removed for DTS-HD Master Audio since the original lossy signal "Y" plays the biggest role (I would wager 90%+ of audible information) in assembling the "lossless" signal. Therefore what you are essentially getting with DTS-HD Master Audio is a signal that is not truly lossless, as the lossy signal is in reality the bulk of the the "lossless" signal and the artifacts in that lossy signal are carried over into the "lossless" signal. Also, despite this DTS-HD Master Audio actually takes up more space than Dolby TrueHD.

Despite its appearance of being inferior sonically on paper, DTS-HD Master Audio's backwards compatibility is a massive booster to that format on Blu-Ray as Dolby TrueHD decoding is *not* mandatory for Blu-Ray players. Because of this, studios will hesitate in using TrueHD on Blu-Ray releases, since there will be no standalones that can actually play the tracks. However, by using a DTS-HD Master Audio track Blu-Ray studios can placate both the customers who want "lossless" sound while ensuring everyone will at least get Standard DTS. This has already happened, with several Fox titles set for DTS-HD Master Audio on Blu-Ray and zero TrueHD titles. On the other hand, with HD DVD we already see many TrueHD titles since TrueHD decoding in the player is mandatory on HD DVD.

So, while DTS offered higher bitrates that generally resulted in better sound on standard DVD, in the high definition realm their codec falls significantly short of Dolby's IMO. Thoughts?

I favour PCM, because it's not going to start a flame war. That said:
I'm afraid this is simply inaccurate. That's exactly what it does. Any competent DSP engineer can design a system which will measure the errors in a signal, with respect to a reference, and generate compensatory data from the difference. Many sorts of digital systems, including all (or virtually all) audio codecs, already do this. In fact, if you look back at some of the pre-digital advanced TV systems, they did just that : downconverted a higher-resolution TV signal (typically something like "720i") into NTSC, and then transmitted the difference in a separate channel, so thay could be "stiched" together at the reciever.

Here's where it gets confusing.

On Blu-ray, despite the lack of required audio decoders in players, the BD format REQUIRES that Dolby True HD tracks also include a standard DD lossy "core" that rests along side the MLP-based lossless signal. Because of this *all BD players can play back a standard DD signal* from BD discs that only officially contain a Dolby True HD track.

So in practice, both DTS-HD and DTHD will, at worst, produce legacy DTS/DD signals (at the max bit rate) from BD discs on even low-featured players.

David, do you have a reference for this? I've seen it mentioned in posts here, but I haven't seen anything to the effect in the white papers I've read (as opposed to DD+, which clearly does). Maybe I haven't been observant enough, but if you could point me to the relevant area, I'd appreciate it.

True. But will the codec be designed that way? And will they produce tracks using that competence?


Cees

Are we all sure that DTS lossless isn't really fully lossless? I knew that it had a lossy core with an extension frame for full lossless quality but assumed since lossless means... well... lossless... that you got real lossless (bit for bit) upon full decoding.

I thought that in order to claim lossless, a bit-for-bit copy of the LPCM master had to be returned. Anything less isn't lossless (though it may be free from compression artifacts... the same way that sample-rate-converting wouldn't be bit-for-bit the same, but wouldn't really be "lossy compression" either.

Does anyone know for certain if the "lossless" term is so flexible that DTS could use it while not returning a bit-for-bit signal? Are we even sure that the codec isn't returning a bit-for-bit copy?

David,

If they use a lossy compression signal plus a separate "difference" signal, they can never "guarantee" that the end-result is bit-to-bit equal again. It will also depend on the decoding logic (e.g. possible timing differences between the two complementative tracks) and the quality with which the lossy track is reconstructed.

It would be too much to ask such an absolute guarantee, but of course, those differences could be made minimal at worst.


Cees

I'm pretty sure that can be guaranteed, with the same degree of certainty that the PCM audio on a CD can be guaranteed to be the same as what was on the master tape, that is, with the occasional uncorrectable burst error. One easy way which comes right to mind is to include checksums for the original PCM with the "correction" datastream. That way, the reconstructed PCM would be compared with the original, and an error flag generated if they didn't match.

Really, the people at dts Inc. would have to be total idiots in order to botch this one up.

Cees

In order for DTS "lossless" to have any validity, the encoding routine would have to compare the "sum" of the lossy core/extension files during the encoding process to ensure that they reproduce the original LPCM data. I would assume that this would be a matter of routine built into the DTS - HD master encoding algorithm (ie, not something that needs to be manually/optionally done or that can run the risk if not being properly validated).

If this isn't the case, then DTS Lossless isn't "Lossless".

Do we have any folks who know the scoop about DTS Lossless and how it works? Would an HDCD-encoded 16/44.1 track, for instance, survive with it's HDCD flag intact after compressing and decompressing with DTS Lossless (as it would with MLP)?

For what its worth;
"Identical to the Studio Master

DTS-HD Master Audio is capable of delivering audio that is a bit-for-bit identical to the studio master. With DTS-HD Master Audio you will experience movies and music exactly as the artist intended.
Bit-for-bit Identical

DTS-HD Master Audio delivers surround audio that is indistinguishable from the original soundtrack or music recording. The new high definition optical discs have far more capacity than standard DVDs. This allowed DTS to develop a surround sound format to deliver surround sound at super high bit rates - up to 24.5 Mbps on Blu-ray discs and 18.0 Mbps on HD-DVDs that are vastly superior to standard DVDs. This bit stream is so "fast" or the transfer rate is so "high" that it can deliver Lossless Audio, a "bit-for-bit" recreation of the original recording. The result is 7.1 channels of audio that are identical to the original studio master. With DTS-HD Master Audio you will experience movies and music exactly as the artist intended."

http://www.dtsonline.com/dts-hd/iden...dio-master.php

To be taken w/a ton of sea salt! ;-)

dts has experience in designing error correcting codecs. Consider dts es discrete.

Wow...here comes these issues again...my previous question arises again: is lossless truly lossless? If we are not DTS and DD designers/insiders, how will we ever know?? Woooooooooooooo! A good spooky halloweeny question...since it is the season...but I'm sitting this round out. People took pride in smashing my pumpkins pretty badly the last time I brought up this debate.

Trick or treat, audiophiles...
Mike

One trick to test codecs like apple lossless is to compress a CD that's encoded for HDCD. Then decompress and burn back to CDR. If the HDCD light comes on the (HDCD enabled) CD player... the compression codec restored the exact bit-for-bit data.

Compressing DTS tracks that were "packed" in red-book CD form and back again is another way to test. DTS had released some music on conventional CD ROMs back in the late 1990's that can be used to test this way.

I hear this will be the subject of the next X-Files movie.

One thing to remember is that for a lossless codec, a maximum bitrate may not correspond to "efficiency" at all. It's a bit budget, and if a piece of music can't fit under that bit budget, compromises must be made.

The MLP track may be a six track 96/24, but that doesn't mean that this rate is constant. Sometimes the surrounds are laid down at 48/24. Sometimes the dynamic range is only 19 or 20 bits. A good engineer will parcel out the bit budget in a way that doesn't affect the musical presentation, but this may not always be possible.

I've taken to ripping my CDs in Apple Lossless, not because my ears are golden, but because it's simply easier than a repeated round of "hmm, that didn't sound right."

Top bitrate is 980 kbs for REM's "Monty got a Raw Deal." Lowest bitrate is Cecilia Bartoli's "Sposa Son Disprezzata": 403 kbs.

Now, it may be that the tracks were mastered in a certain way, but Pink Floyd's "Money" is 955 kbs, and another track from the same album,"The Great Gig in the Sky" is only 732 kbs. "Speak to Me" is 553 kbs, but the comparison is rather unfair...

Jeremy,

it's even simpler than that. No engineer has to tweak anything. You're right that lossless is a variable bit-rate... but there's no user-selectability: The codec compresses and alocates bits according to the most effcient packing that it can do.

If enough bandwidth wasn't available given the program material and resolution, then the lossless process would not be possible (ie, no "compromise" can be made at the lossless compression level... it's either lossless or it's not).

The more resolution and more complex the waveforms the higher the bandwidth demands, all things being equal. So if a 24/192 7.1 channel doesn't have enough bandwidth, then sure, the engineer might make a "compromise" in the sense that he might choose to use a downconverted master resolution to, say 24/96 or 20/48 etc. and then try lossless compression again and see if it fits. However, these aren't compromises of compression, they represent different LPCM resolutions for the source that's being encoded.

certain studios were known to prefilter their masters before converting them into mpeg2....

Hey Michael,

I guess our concerns were legit. This scenario reflects exactly what I wrote about on the previous thread (more precisely posts 4 and 7)

http://www.hometheaterforum.com/htf/...08#post2930608

For me PCM 48/24 (or whatever the original master sampling rate is) is the grand goal. From the technical standpoint, I know a lossless compression is more efficient, but using an economics term, since information is asymmetric, sometimes we can't know for sure how things work. So in this case, why not go with PCM multichannel? There should be absolutely no compromise in video quality if it's used along side with VC-1 (preferably, because it's been developed more closely with studio feedback) or AVC on a BD-50 (before format wars jump in, I'm supporting both, but I believe BD has higher potential).

So I give my kudos to Disney for being the only studio up to now to provide lossless PCM with the original master's sampling rate. Truly master quality indeed ; )

I thought DTS Master Audio worked as basically two separate tracks: one lossy and one lossless piggy backing on top. The lossless soundtrack is the extension, the lossy is the core. If a DTS decoder can't do the extension portion, it ignores it and plays the core. If the decoder can do Master Audio lossless, it plays the extension instead rather than a combination of the two. That's what I gathered on a discussion on AVS Forum where it was mentioned they were really two separate tracks, but combined for ease of backwards compatibility. That was why DTS Master Audio was not as bitrate efficient as Dolby TrueHD. If someone knows if this indeed in error, please feel free to correct this data.

Dan