On the subject of sound in the brave new era of Digital Cinema.

Having been asked a number of times about this in recent weeks, I have decided to compile a summary. I’ve talked a little about the image in the past, but I seem to remember Dolby saying “sound is half the picture”.

As always, this applies to any cinema using professional digital projection systems – these meet a number of criteria including content protection methods (the days of distributors trusting exhibitors are long past). This post, and the information presented, does not apply to any cinema in specific.

A little bit of history to put this in to context first. When cinemas used 35mm film, (for the most part) sound was stored on the film. Wikipedia has a detailed page about it. Modern cinemas were able to derive digital sound from 35mm film by using one of the proprietary digital cinema sound systems, common names such as Dolby Digital, Sony Digital Dynamic Sound, and Digital Theater Systems.
What all of these had in common was a “data reduction” system, to reduce the storage capacity requirements to fit in the limited space available on 35mm film. Dolby used a system called AC3 (which lives on in the world of DVDs), Sony used a system called ATRAC, and DTS used a system called APTX-100 (sadly there is limited information about this).
The human ear is a sensitive instrument, and there are subtle acoustic differences to how these systems encode and decode sound. In the world of home cinema, the difference between the consumer versions of Dolby and DTS is pronounced and clear to me.

I am of the opinion that the introduction of digital sound to 35mm film was a great technological achievement, and most of the time it enhanced the viewer’s experience. In a very small minority of times (so I hear), it caused some problems for cinemas. I believe these were actually caused by faulty film printers.

Moving forward.

If you look at the Digital Cinema Initiative specification, section 3.3 “Audio Specification” says: Bit Depth
The bit depth shall be 24 bits per sample. DSM Audio Material having other bit depths shall be justified to the most significant bit per [AES3-2003 Section 4.1.1]. Sample Rate
Irrespective of the associated image frame rate, the audio sample rate shall be either forty-eight or ninety-six thousand samples per second per channel, commonly expressed as 48.000 or 96.000 kHz. At 24 FPS playback, there are exactly 2,000 audio samples per frame for 48.000 kHz and exactly 4,000 audio samples per frame for 96.000 kHz. At 48 FPS playback, there are exactly 1,000 audio samples per frame for 48.000 kHz and exactly 2,000 audio samples per frame for 96.000 kHz.
A theater playback system shall have the capability of performing sample rate conversion as needed. Channel Count
The delivered digital audio, contained within the Digital Cinema Package (DCP), shall support a channel count of sixteen full-bandwidth channels.

In plain English, this means that the sound is never “compressed” (data-reduced), and you’ll notice that digital cinemas never advertise any kind of digital sound format because of this.
It doesn’t matter if you have a Dolby-branded sound processor (for example), while showing DCinema (DCI) presentations there is no decoding taking place. There are plenty of other sound processors, from other manufacturers, which will fill the same role in the same situation.

The only sound processing that does occur is equalisation to compensate for the acoustic nature of the auditorium, and (in some cases) down-mixing of channels where discrete amplifiers and speakers are not present. Equalisation (for historical reasons called B-Chain Equalisation) is a requirement for all cinemas, film and digital alike.

The standard CD audio disc, provides two channels of 16 bit linear PCM sound, with 44100 samples per second.

The DCI specification insists on 24 bit linear PCM, this gives a greater dynamic range from very quiet to very loud.

The DCI specification has two options for the number of samples per second, either 48000 or 96000. This (in theory) affects the highest frequency of sound which may be reproduced. Mr Nyquist says that a sample frequency of 48000 samples per second (a.k.a. 48kHz) means that the highest frequency which will be reproduced is 24kHz. This figure is about at the limit of human hearing.
Having the option of 96kHz sound is complete over-kill, and I doubt there are many amplifiers and speakers in cinemas which will reproduce anything (Nyquist) above 30kHz never mind the theoretical limit of 48kHz.

Finally, the DCI specification allows for a maximum of 16 channels of sound, and section 3.3.3. gives the mappings for where speakers are located relative to the audience. In practice, the traditional 5.1 layout is very common, particularly with cinemas which have converted from 35mm to digital.
When someone asks me what the sound format is, I tend to say “Linear PCM”. Which is strictly accurate, although most people will have heard this (in the context of audio CDs) but don’t know what it means.

(10/9/2010) I have deleted the last paragraph, which was interesting by its self, but didn’t add anything to the central discussion.


Tags: , ,

One Response to “On the subject of sound in the brave new era of Digital Cinema.”

  1. barrowbiker Says:

    This entry is attracting a number of spam comments about “streaming movie” sites. Despite what a minority of people have told me – I am of the opinion that streaming a movie (particularly if it not yet in cinemas) is theft. No matter how you dress it up, and it makes no difference if no local copy is saved on your computer.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: