|  IP: Logged

 |  IP: Logged

You may be right about the frequency of SDDS tracks showing up on prints decreasing. I can't really give you exact numbers, obviously, I was just relating my own experiences over the past few years. You are, of course, correct about SR-D showing up most often!

Isn't that 30:1 compression ratio on SDDS only for the secondary / back-up digital track Gordon? Or am I just not understanding something here?

It is not the "amount" of compression that is relevant anyway, but how it is used. The real issue is that DTS uses a system (called APTX) in which the data is fully recovered (well, 99%), and is fully intact - unaltered from the orginal. There are virtually no losses and the audio is equivalent to CD audio.

However Dolby (and to a lesser extent, SDDS) use systems based on Dolby's AC3 compression scheme which discard (forever!) large amounts of audio data, particularly when all 5.1 channels are carrying lots of information at the same time. These systems use what Dolby refers to as "psycho-acoustic masking" which is a method of data compression supposedly modelled on the human ear, and it's tendancy to not hear certain quite sounds in the presence of other much louder sounds. So the quite stuff (that you're not supposed to be able to hear) is just thrown away. The more channels that are working at the same time, the more audio that is discarded.

However anyone with a half way decent working set of ears would have to conclude after listening to any SRD track that the human ear that AC3 is modelled on was probably in excess of 70 years old!

SRD does sound quite acceptable when only the centre channel is working, but add stereo music and effects in the surrounds at the same time and the whole thing starts to really come apart. Anyone disgree - (all of those deaf heads out there need not reply).

------------------

The big mistake that the THX people make is placing their four analyser microphones all over the damn shop. They try to create an average response in very last nook and cranny of the auditorium, and the fact is you just can't do it, and this method just creates more problems than it solves.

We have found over and over, that placing the four micphones in a tilted diamond pattern which is restricted to the main seating area (two thirds back where most people sit) works every time and produces accurate highly repeatable results. You go back in six months and set up the mics again and the response is almost identical.

your description of Dolby Digital sound rather strange to me.
When listing to good mixes like Matrix or Fight Club in good sounding cinemas, I have not all the feeling of "monophonic sound splitted to 5.1". Or did I miss something of your points.
In fact I have no real problems with Dolby Digital sound quality of these movies. This at least is true for the original version. The German dubbed versions have inferior sound quality as usual.

For clarification of the discussion in this tread: Here are the parameters for the digital film formats, given in the sequence Dolby Digital / DTS / SDDS.

Resolution: 18 / 16 / 18 Bit.
Sample Rate: 48 / 44.1 / 44.1 kHz.
Data reduction: 13.5:1 / 4:1 / 5:1.

Discrete recorded channels:
Dolby: 5.1
DTS: 5.0, the subwoofer is recorded in the surround channels LS, RS with Xover at 80 Hz
SDDS: 12.0, subwoofer is fullrange. There are 4 failsafe channels C, SW, Mix of L+LC+LS, Mix of R+RC+RS.

Data Rate: 320 / about 880 / about 1900 kBit/s.
(for DTS and SDDS the data rate is calculated from the other data as I did not find a specification).

For SDDS we should consider that usual 5.1 mixes only uses 5.05/12 of the data rate (=about 800 kBit/s, the subwoofer only requires 1/200 of the bandwith).
So the data rate for SDDS is not so impressiv anymore.

Physical formats on the print:

Dolby Digital
Size of a bit: 33 um (micro meter).
78 bits by 78 bits between each perforation, giving
Physical data rate: 584 kBit/s, the difference to the data rate 320 kBit/s is used for syncs and error correction

SDDS
Size of a bit: 24 um
Physical data rate: there we have a mystery.
On one side I found the spec 4000 kBit/s.
On the other side I found the specification that 64 bits minus 6 for snyc are traveling on each side of the film. This calculates to a total of about 2200 kBit/s.
Where are the other bits stored?

The smaller size of the bits might be an other reason for earlier wearing out of SDDS tracks.

Data reduction of 3 digital formats is not lossless. So it is important to know/hear the defects during the mix as (it is necessary for matrixed analogue mixes).
Dolby: the DS10 encoder is provided to the mixing studio, so the studio has control over the encoded/decoded sound.
DTS: external encoding, the DTS-Tower provided to the studio deals with the different record format (subwoofer in the surround channels).
SDDS: external encoding

After playing, a rep from Dolby gave a short talk/QA session about AC3. He made a specific point to say that; Dolby, Inc. was *not* saying AC3 was the best at everything.

He said that Dolby's offical position is that; AC3 is the best sounding process using only 256kb/s. Dolby acknowledges there are better compression algorithms. However, bandwidth is limited in most applications- there's only so much room on a DVD, HDTV signal, modem, or space between sprocket holes. Dolby wanted *one* process that would work for all these applications.

If you've ever done any network administration, or worked on engineering projects requiring "real-time" analysis, etc. you quickly see that there is always a struggle for bandwidth.

So, while not perfect, I think AC3 was a "real-world" solution that does quite well for the task(s) it was designed for.

Besides, I don't feel that super-high-quality audio is obtainable in a theater auditorium.

I think that the highest compressed soundtrack (AC3) still has more quality than the sound system's ablility to play back. The reasons are; the speakers and the acoustics.

Speakers for theaters need to be designed to "fill" a large area, and auditorium acoustics are difficult to control (unlike a home system where the smaller speakers can be made for more accurate reproduction, and the room acoustics will not change significantly after EQ'ing.)

Look at what happens in a theater; after putting up the speakers, we put a "low-pass filter" (the screen) in front of them. Then, we EQ an empty auditorium, even though it can get filled with people. To cover the widest area, some seats usually need to be "left out" of the HF speaker's coverage area, resulting in those patrons hearing poor HF quality.

So, my point is that the speakers and acoustics will degrade the audio signal enough such that (I feel) compression rates do not matter. Let's start working on picture quality.

just to clarify. The SDDS System, according to the SDDS Recorder System's product information manual states the following data:

Sample frequency 44.1 kHZ (could adopt any other rate by high quality sample rate converter)
Dynamic range 90 dB
Number of channels 8
Compression rate 1:5 (ATRAC 100)
Recording bit rate 4 Mbit per second
The system is discrete 12 full range channels (8 + 4 backup), which in return will deliver around 333 kbps per channel.

It's rate per channel is higher than AC3's shared 256 kbps used for 5.1 channels in a bit bucketing technology.
This is why I tend to state as a "shared mono system".
The compression rate in practise is somewhat 4.5 :1 and is identical with ATPX 100's. DTS states 4:1, whereas SDDS states 5:1, whatever mathmatical rounding is applied.
AC3 could also be 15.8 : 1 with very saturated mixes.
We also have to keep in mind, the versions of the compression systems on all formats are rather 'old', as the date back to about 1990, in microelectronic timeschedules, and that some consumer product versions rely on more advanced algorithms. Correct, DTS is totally different, and no prediction is used.

Both systems on film (SRD and SDDS) rely on derivates of the Dolby proprietory BCIRC (Block cross interleave Reed-Solomon) error correction algorithm.