DTS-CD DTS - "lossy"?

[StarTrek1701]

Sorry for my ignorance, but...

I have read a number of threads where DTS is called "lossy" and DVD-A is lossless (I hope I got that right). So, what is being lost? Dynamic range? Frequency range? And how much is being lost.

For example, let's say it is frequency that is being lost. If it is off of the bottom end of the scale, I can see that as being a really big concern. But if it is off of the top end, the affect would be negligible. The current standard is 20 to 20,000. But, a Q8 deck will only playback between 30 to 16,000 (my Quad reel stops at 18,000). So, I could afford to loose 2,000 or 3,000 off of the top without any detrimental affect to the music I am transferring.

Now, all of the above is a guess. So, some info would be greatly appreciated.

Thanks.

Glen

 

[winopener]

Lossy=throw away musical data. It is not a matter of bit depth (16,20,24) or frequency range (44100, 96000...) but on compression scheme (MP3 is lossy, FLAC is not). If you encode to a lossy codec, whatever it is, you can't have back the original file as it was, while if you encode to a lossless codec (flac, mlp, ape...) you can have back the original file as it was.

 

[MidiMagic]

Most compression schemes remove what seems to the software to be redundant information. This is often in the form of harmonics and sounds too low to hear. Some of that is often the ambiance the quad listener often wants to hear.

 

[ar surround]

The recent release of Tull's The Broadsword and the Beast DVD still relies on lossy 96/24 DTS...even in this age of lossless blu-ray being the common / preferred format. I've always wondered, what is it that the pair of typical QQ member's ears notices on the lossless stream that is missing from the lossy
stream?

Has anyone had a chance to make an exact comparison of the lossless and DTS lossy streams on any title that is out there? If so, what did you notice?

Note that I'm specifically interested in 96/24 DTS, not streamed Atmos with its tiny file sizes. Also, don't bring up Dolby Digital because it definitely has a 'grizzly' sound to it that is quite obvious.

@sjcorne Jonathan, you are one of our resident gurus on this stuff.

I can only surmise that my old ears might notice a very slight difference. But in a double blind test? Perhaps I might even choose the lossy version. Who Knows?

 

[fredblue]

in a double blind test of the exact same material, mastered identically and level matched, i doubt too many (myself very much included) would be able to discern between Lossless Surround and DTS Surround at 96/24

 

[LuvMyQuad]

I always wondered if lossy DTS 24/96 would be worse, better or equivalent to the lossless 24/48 streams found on Blu-ray.

 

[JediJoker]

I always wondered if lossy DTS 24/96 would be worse, better or equivalent to the lossless 24/48 streams found on Blu-ray.

Worse. But audibly worse? Probably not.

 

[par4ken]

For the most part lossy DTS can still sound very good. DTS even claimed that they used compression to improve the sound! A 16-bit CD encoded DTS produces a decoded 20-bit output. The difference is that the lossy files have a slight grittiness to the sound that a lossless file doesn't. The same is true of MP3 compared to wav or flac. You just can't throw away bits of sonic information and expect to get the same audible result.

There is no way that a lossy file could ever sound better than a lossless one! Depending on the deficiencies in your audio system the differences might not be audible to everyone.

 

[Owen Smith]

DTS 24/96 is a dumb concept. This is lossy compression, so why waste bits conveying frequencies that only bats and dogs can hear? Those bits would be better used to throw away less of the information humans can actually hear.

 

[par4ken]

A lot of thought went into the development of DTS. A lot of very carefully controlled listening tests were done as well. Here is a copy of the "White Paper", It contains far more information than I care to try to fully digest! They like to push the idea that linear PCM is not a very efficient way to code audio.

 

[Owen Smith]

A lot of thought went into the development of DTS. A lot of very carefully controlled listening tests were done as well. Here is a copy of the "White Paper", It contains far more information than I care to try to fully digest! They like to push the idea that linear PCM is not a very efficient way to code audio.

I'm aware a lot of thought went into DTS, but lossy 24/96 was not part of that original development.

I agree that LPCM is not an efficient way to code audio. That's why we now have things like MLP (which compresses silence poorly), FLAC, ALAC, and DTS HD Master Audio lossless compression. Heck FLAC can almost get 5.1 channels into the original full rate DTS bit rate, on average FLAC can get 4 channels of 16/44.1 into 1.5 megabits.

 

[par4ken]

I'm aware a lot of thought went into DTS, but lossy 24/96 was not part of that original development.

Did you look at the white paper? Quote from the Introduction.

"Coherent Acoustics is a digital audio compression algorithm designed for both professional and consumer applications. The algorithm is highly flexible, operating over multiple channels of audio, at sampling rates up to 192 kHz, with up to 24-bit resolution. This article outlines the overall design objectives of the algorithm, and describes the principal coding strategies employed to meet these goals. Specific functional components of the encoding and decoding algorithms are outlined in more detail, and the feature set of the coding system is summarized. Finally, the objective performance of the algorithm is illustrated using some simple test signals."

I agree with you, I personally would prefer a lossless codec!

 

[barfle]

While my 76-year-old ears miss a lot that my 25-year-old ears used to hear, everyone whom I’ve demonstrated my system to is wowed by the sound. So someone with less ringing in their ears might well hear deficiencies, I think most of my DTS recordings sound pretty good, which has always been my goal. “Pretty good” mostly means that it makes ME happy. I have a handful of MP3s that I downloaded in the 00s that sound like crap, so my ears aren’t totally tin, but I believe a younger critical ear could hear some issues that I don’t.

 

[barfle]

Did you look at the white paper? Quote from the Introduction.

"Coherent Acoustics is a digital audio compression algorithm designed for both professional and consumer applications. The algorithm is highly flexible, operating over multiple channels of audio, at sampling rates up to 192 kHz, with up to 24-bit resolution. This article outlines the overall design objectives of the algorithm, and describes the principal coding strategies employed to meet these goals. Specific functional components of the encoding and decoding algorithms are outlined in more detail, and the feature set of the coding system is summarized. Finally, the objective performance of the algorithm is illustrated using some simple test signals."

I agree with you, I personally would prefer a lossless codec!

While I just saved the paper (so I haven’t read it yet), the idea of lossy codecs is intriguing. I’m pretty familiar with MPEG encoding (lots of redundant information frame-to-frame in movies and TV), I can’t say I have much understanding of how audio compression works, so further study is warranted. the paper isn’t all that long, so I’ll be giving it reasonable consideration later this evening.

That being said, in a perfect world, lossless can’t help but be superior to lossy, but the difference may not be audible to most listeners. I believe that was the goal of most codec developers, although in the old days of 1200 bps modems, data rate with AM crystal set radio quality may well have been the design parameters.

 

[ar surround]

While my 76-year-old ears miss a lot that my 25-year-old ears used to hear, everyone whom I’ve demonstrated my system to is wowed by the sound. So someone with less ringing in their ears might well hear deficiencies, I think most of my DTS recordings sound pretty good, which has always been my goal. “Pretty good” mostly means that it makes ME happy. I have a handful of MP3s that I downloaded in the 00s that sound like crap, so my ears aren’t totally tin, but I believe a younger critical ear could hear some issues that I don’t.

Our eyes probably ain't so great either...

 

[Sonik Wiz]

Another fine example of how threads never die on QQ. Midi's post was 13 years ago & still there are questions & worthwhile discussion on the quality of DTS. I love it!

To @ar surround I can't answer your question specifically. But I do have my own learning experience about the transparency of DTS compared to lossless.

About 15 years ago I sat about upmixing all 14 DVD's of Chisato Moritaka. These were a mix of music video's & live performances. Being DVD ,of course, standard def but they all had 16/48 LPCM audio. The work flow was pretty straight forward: demux audio/video (no transcoding) > upmix audio on AA 3 > export finished project as 6 independent WAV files > encode to DTS > remux with menu/chapters in DVDLab Pro > burn disc.

When working on something like this I have a wireless keyboard & mouse & I can sit in my listening sweet spot and view AA 3 on an 8' screen. So I mix it to sound just right for me & those conditions. I can also evaluate quality of the DTS encode to the lossless files the same way. With the multi-track view of AA 3 on the screen I cue it up & plop a disc in the Oppo. The Oppo outputs analog audio to my Anthem pre-pro & there's no room correction or the such to alter it. Level matching is very easy between sources.

So with a push of the button on my remote I can switch between DTS audio & lossless audio from the PC. There's one & only one way I can tell the diff between lossy & lossles & that's the DTS always had the slighly less bass to it. Not much... 1>2 dB reduction in the low end. I never figured out if it was a change in LFE ch or just genera bass reduction. At any rate all else sounded identical. Attacks & timbre on conga drums, the schifff on saxaphones reeds, the ting of a triangle, no matter how I tried to hear discrepancies I could not. Oh I guess I'll metion this DTS was encoded at the full bit rate of 1509.75 kbit/s. The truly geeky among us will know that after the launch of DTS DVD's the bit rate was cut in half for more digital room allowing for previews, add'l languages, etc. At full data rate it's a great performer.

If I do an audio only project now days I will just save to FLAC. But FLAC is not compatible with video standards (as far as I know) so DTS is still a good option.

 

[Bill B]

While my 76-year-old ears miss a lot that my 25-year-old ears used to hear, everyone whom I’ve demonstrated my system to is wowed by the sound. So someone with less ringing in their ears might well hear deficiencies, I think most of my DTS recordings sound pretty good, which has always been my goal. “Pretty good” mostly means that it makes ME happy. I have a handful of MP3s that I downloaded in the 00s that sound like crap, so my ears aren’t totally tin, but I believe a younger critical ear could hear some issues that I don’t.

In the end being happy is all that matters. I always strive to get the best sound but try to never lose sight that it isn't what's most important. It's the music and the songs. When I am washing my truck with a lousy $100 boombox reading a flash drive with a pile of crappy mp3s playing, nothing sounds better. Furthermore the additional enjoyment I would get out of it, at that moment, if it were a better system is minimal.

 

[marpow]

Great question and great answers.

I probably can't hear the difference. As a consumer of both Lossy and Lossless music, I have heard outstanding DTS Lossy and not so lossless releases.

As a hobbyist of music, I do notice more easily Dynamic Range compression. As an experiment I ordered some original, 1990's CD's to compare against my later years remastered CD's and the earlier CD's with high DR numbers sound much better than the later years remasters CD's.
For a moment I thought about search and replace, but that would be foolishly expensive, but nice to know I could if I wanted to.

I totally agree with @Bill B. I was in Jamaica about 15 years ago, no CD's, boom box or anything. Cranked up my iPhone, turned on my library of MP3's listening to these little 1/4" phone speakers, while putting the iPhone in a bowl, and we had a great time, playing cards and listening to music.

 

[PurpleMoustache]

There’s quite a few good tests out there that stream lossless and lossy audio to you as an ABX test, here’s one I usually link when this comes up.
I, for one, having gone through that test cannot tell the difference. That doesn’t stop me from streaming/downloading/listening to lossless. I might as well use the best source as possible, even if I cannot tell the difference

 

[jimfisheye]

...

Has anyone had a chance to make an exact comparison of the lossless and DTS lossy streams on any title that is out there? If so, what did you notice?

...

I did just that with the bluray vs DVDV releases of Tull Aqualung. It turned out that the quad program was the same source for both formats. (Other program on the discs had different mastering.)

I can usually hear something A/B'ing between wav and 320k mp3. Sometimes it's subtle or impossible to tell. Turns out that this format even though lossy is still robust! I've heard tape change thing far more.

I couldn't hear any difference A/B'ing between the wav and dts2496 copies.
Nulling them of course produced results. Audable results too just like a null test with a 320k mp3 vs the original. So there's loss there alright! I'm usually sensitive to generation loss artifacts but I guess this format is working for me. Considering all the gross stuff and volume war mastering and novelty releases... shit, I'll take this with a smile!

For some other examples.
Sample rate conversion for example. (With either SOX or r8brain)
Null tests on pre vs post usually null down to -90 or -100db and you hear silence at non-dangerous monitoring levels. So the lossy conversions are plainly more evident than a sample rate conversion.

Atmos is lossy like dts2496. That surprised me a little since it's core is Dolby TrueHD which by itself is lossless. Again, can't hear a difference A/B'ing by ear but a null test pre vs. post reveals an audible difference result much like with dts2496.

Disclaimers:
I don't have the dts encoder suite so I can't run a strict test with my own audio. Looks for all the world that those Tull test cases were the same source though.

I do have the Dolby Atmos encoder suite and have compared with my own audio files.

Again, no complaints for either one! I'd be none the wiser just going by ear.

The mix is a big deal. Hopefully obvious!
Non destructive mastering (that means no volume war or bionic treble blast) is a big deal.
These delivery formats are solid.