Sound

Digital Delay For Application in Surround Sound

1.0 INTRODUCTION

1.1 MULTI-CHANNEL REPRODUCTION

1.1.1 Hafler Surround Sound

David Hafler [2] first proposed a system that would both encode and decode information on existing two channel stereo whilst still being compatible with normal recording and playback techniques.

If information is encoded onto each channel with equal amplitude but opposite phase it can be decoded by simply taking the difference between the two. Similarly if the information is encoded with equal amplitude and the same phase in both channels, then it can be decoded by summing them both.

For a recording made using coincident-microphone techniques a signal could be derived between the loudspeakers outputs on reproduction which corresponded to the output of a sideways-facing figure of eight microphone.

This means that the illusions of sound sources could be enhanced beyond the space between the main loudspeakers if this signal is replayed through extra loudspeakers which are placed to the sides of the listener [3].

This method works particularly well on sources such as live classical music recordings which are recorded in the natural surroundings of concert halls. Sound energy from the orchestra will be reflected all around the hall and eventually what arrives back at the microphone will be out of phase information.

One extra speaker will suffice (if it is placed to the rear of the listener), but if two are used, with one of the two wired in anti-phase, then a wider image can be created.

For the ideal situation these extra speakers should be placed to the rear of the listener, further apart laterally than the two main speakers and slightly higher.

If the rear speakers signal is delayed with respect to the front main speakers, then an effective shift can be obtained in the rear signal giving greater depth to the overall sound experience.

As an approximation sound will travel three Metres every milli-second, and so a delay of twenty milli-seconds will create a rear-ward image shift of approximately sixty Metres.

The information contained in the signal for the rear speakers will in general be lower in level than that of the main speakers and external processing will be required to alter this, subject to the listening environment and the listeners taste.

The frequency content of the signal will be predominantly in the mid-range, typically 200Hz to 6kHz and so it is not necessary to have speakers capable of reproducing the full audio bandwidth. In fact ensuring that the higher frequencies are filtered out may be advantageous, as this would ensure that important directional information appears only to come from the front, otherwise the result may appear unnatural and disturbing to the listener.

1.1.2 Quadraphonics

The first commercial attempt to go beyond stereo reproduction was in the early to mid 1970's when the concept of quadraphonic sound was unleashed on the general public.

Quadraphonic sound never really caught on and there were a number of reasons why this was so.

The first problem was that there were a number of different systems on the market which were largely incompatible with each other. The main three being: CBS's SQ, Sansui's QS and JVC's CD-4. The first two were matrixed systems, and involved processing four channel sound information into two channels. The latter one, CD-4, involved having two basebands of stereo and modulating two more onto a high frequency carrier [4].

All of the systems suffered from technical difficulties: SQ and QS because the correct way of handling the matrix operations was for more complicated than it appeared, and CD4 because of the need to record and recover very high frequencies from a vinyl disc records [3].

Despite all the incompatibility problems there was also the problem of physically placing four speakers in a listening environment and still being able to sit in a prime listening position. It is said that American houses tend not to have door entrances in the corners of rooms unlike their British counterparts, where the placement of a these extra speakers always seemed to end up causing problems [3].

1.1.3 Ambisonics

The early 1970's saw much work done by Peter Fellgett and especially Michael Gerzon [5] who through the NRDC (National Research Development Corporation) [6] developed the necessary mathematical theory which allowed a rational system of multichannel surround sound. This system is known as AMBISONICS. It is able to reproduce the directionality of indirect reverberant sounds, as well as direct sources.

All rights to ambisonics are now owned by Nimbus Records Ltd [7] giving them the rights to all patents for ambisonics, except those for microphones. Nimbus have been producing C.D.s in ambisonics for a number of years. Live concerts are recorded on a soundfield microphone, or equivalent, positioned approximately 3 Metres behind the conductor. Such a microphone has the capability to pick up sound in all directions. It has a figure of eight directionality forward facing with a gain of +1.414 for sound straight ahead, zero at the sides, and -1.414 behind (the output of which is termed X). A similar arrangement for the side (output Y), and an omnidirectional response (W). If height information is to be acquired also then a figure of eight direction pattern is used (Z).
For derived responses including height, the term periphonic is used [5]. Essentially only three channels are needed for 360° surround sound in the horizontal plane.

Four different formats exist in the encoding and decoding of ambisonic information [5], [6] & [8].

A - FORMAT: Synthesis of signals representing both the sound wave and direction (e.g. in a four channel system Lf = left front, Rf = right front, Lb = Left back, and Rb = right back).

B - FORMAT: This is the matrixed information from the A format, or the output of a soundfield mic.

X = 1/2(-Lb + Lf + Rf - Rb)

W = 1/2(Lb + Lf + Rf + Rb)

Y = 1/2(Lb + Lf - Rf - Rb)

Z = 1/2(-Lb + Lf - Rf + Rb) (If height information is required)

C - FORMAT: An encoded version of the B format for consumer use. Predominantly used is the UNIVERSAL H.J. (U.H.J) system which utilises two signals that correspond to the left and right of conventional stereo, but contain all the information for full surround sound extraction.

D - FORMAT: The C format decoding technique.

The encoding and decoding processes use both amplitude and phase matrices to take account of the fact that the brain interprets sources differently at different frequencies.
At low frequencies the perceived intensity is the sum of pressures, where as at higher frequencies it is the sum of energies [9].

1.1.4 Film Surround Sound

One of the first films to attempt an all around sound was the Walt Disney's 1940 film "Fantasia", which had discrete six-channel sound [10], as did the cinerama series of movies in the 1950's [11].
The main pioneers of cinema surround sound is the company Dolby who first commercially exploited the idea, when in 1975 Dolby Stereo was introduced.

Dolby Surround Sound Encoding

When a film is encoded with surround sound information the centre channel is attenuated by 3dB. It is then combined in equal proportions to the unaltered left and right channel information. For the surround signal it too is attenuated by 3dB then band-limited at 100Hz and 7kHz before passing to a modified Dolby B noise reduction encoder. It then passes to a +90° wideband phase-shift network before being added to the left-plus-centre signal, and to a -90° phase shifter before adding to the right-plus-centre signal. The two composite signals are then recorded onto the film [12], [13].

FIG 1.0 Dolby Surround Sound Encoding

Dolby Surround Sound Decoding

At this stage it is important to differentiate between the different types of Dolby sound, beyond the scope of noise reduction circuits used in optical and magnetic sound reproduction, Dolby sound can be defined into two main sections:

1) Dolby Surround - This deals with home cinema entertainment systems, for which the equipment is made by licensed consumer electronic manufacturers [14]. It is most closely related to the SQ matrix of the 1970's quadraphonic systems.
Dolby license two types of products for home cinema entertainment. One is the DOLBY SURROUND DECODER which is a passive device and simply derives the surround information from the two channel stereo input. It also has a time delay, bandwidth limitations and noise reduction.
The other is the DOLBY SURROUND PRO LOGIC DECODER which is an active device and uses a variable matrix with additional enhancement to the directional information to improve separation between all channels. Unlike the passive decoder it can run with a centre speaker for locking dialogue to the screen more precisely [13], [15].

2) Dolby Stereo - This concerns the cinema side of surround sound, and of which there are four main systems. The newest is the Dolby Stereo SR D which incorporates a six-channel digital optical soundtrack in addition to a four-channel SR analogue track on the same 35mm prints [16].

FIG 1.1 Dolby Film Sound Classifications

The decoding principle is essentially the same for cinema and home cinema systems. After encountering input buffers the two signals are fed to a matrix which will derive the left and right channels, surround and centre if required. The surround signal first encounters a delay. The next stage is a 7kHz lowpass filter and then a modified Dolby B noise reduction circuit. The latter two circuits serve to quietening delay line and optical sound track noise and to reduce sibilant bleed-through of centre encoded dialogue due to relative phase and amplitude errors in the original two signals. These are most likely to occur from slight adjustment and positioning errors of the two recording light valves and the two playback photocells.
For cinema surround sound an additional high-pass filter set to about 100Hz is applied to protect the surround speakers, which of necessity are much smaller than the front speakers [10], [17].

The time delay is principally to enforce the Haas effect. If the surround speakers are delayed with respect to the front speakers then the sound will appear more evident from the front. The delay time is typically within the region of 20 - 50 milli-seconds, depending on the environment. This may also help to reinforce certain sounds because of the intelligibility aspect of early reflections within this period of time.

THX

THX is decoding system which relies on using material which has been encoded with Dolby surround sound [11], [12].
There are two types: i) LUCASFILM THX and ii) HOME THX.

LucasFilm THX affects the presentation of movies at several points.

For the home, a THX-certified processor starts with Dolby Pro-Logic and adds:
* Surround channel decorrelation - a digital pitch shift is used to make the (mono) surround signal "different" in the left and right surround channels.
* Re-equalization of the front channels, to make the movie mix seem less "bright" in the home.
* "Timbre Matching" - an equalization applied to the surround channel to make effects sound consistent when panned between front and surround speakers.

THX specifies front speakers with a reduced vertical dispersion (to minimize ceiling reflections) and two side-mounted surround speakers configured for dipole radiation.
THX also recommends equalization for the L-C-R channels. A THX equalizer will have 1/3-octave bands from 80 to 800 Hz, implemented as "interpolating constant-Q" circuits, and parametric equalization above 1000Hz and for the subwoofer channel.

"THX" is a LucasFilm trademark for several things, two of which are directly related to home surround:

1. "THX Theatre" - THX is a certification process. Theatres bearing the logo are periodically tested to ensure that they meet LucasFilm standards for audio environment and playback of surround-encoded film.
2. "Re-recorded in a THX theatre" - THX logos on films and recordings indicate that the final Dolby MP-compatible mixdown was done with the recording console and engineer located in an actual THX-certified theatre. This is intended to ensure that the film audio will playback in a consistent and predictable manner in all THX theatres (and in homes equipped with THX certified components).

3. THX crossover - LucasFilm lists recommended audio components for THX theatres. They also make a crossover, bearing the THX brand, which is only used in actual motion picture theatres.
4. Home THX - LucasFilm has a testing and certification process for home audio equipment. Those models which are submitted by the maker, and pass the tests, may exhibit the branding. THX branded equipment provides the promise of effective home theatre, but can still sound poor if improperly set up and calibrated. Some THX-branded equipment includes dealer installation and adjustment. For amplifiers, THX merely provides an assurance of high quality.
5. THX certified surround decoders, equalizers, main speakers and surround speakers, on the other hand, must provide specific THX required functions, as well as high general quality [11].

1.2 STEREO IMAGE WIDENING TECHNIQUES

1.2.1 Holophonics

A technique which was used in the early 1980's was a system called holophonics. It was principally a binaural recording process using a dummy head. The effects, were by their very nature, only at there best on headphones. There were two notable albums which used this technique which were attributed on the albums to the pioneer Zuccarelli Labs Ltd [18], [19].
The system had favourable reviews, but never really caught on in the commercial sense. This is probably because of the requirement to use headphones in order to obtain the best effects.

1.2.2 Q Sound and Roland Sound Space (R.S.S.)

These are both stereo widening techniques that rely on complex phase transformations that give the illusion that a source is beyond the extremes of the left or right speakers.
Because of the complex phase transformations the B.B.C. discovered that material encoded with Q Sound would give very poor mono-compatibility on their F.M. transmissions and so it was effectively banned from radio broadcasts [20].

1.3 AUTHOR CONTRIBUTION

From previous experimentation it was discovered that by using the Hafler technique for deriving surround sound a new listening experience could be obtained from ordinary stereo sources. What was needed was a dedicated processor in order to achieve this.
The major obstacle to overcome in creating a dedicated processor which would carry out this operation, is the requirement for a delay. Although there are a number of semiconductor chips for sale which have the ability to delay audio signals (known as bucket-brigade chips) they tend to be:

i) rather noisy,
ii) have a restricted bandwidth,
iii) have a small dynamic range,
iv) can only create a relatively small delay time.

Despite these limitations popular electronics magazines have published designs for such surround sound systems using this technique to generate a delay [21].

It was therefore because of the aforementioned reasons that the decision was made to create the delay digitally.
This had the following advantages:

i) the full audio bandwidth could be catered for.
ii) the signal to noise ratio would be a lot higher - provided sufficient bit resolution was used.
iii) the system would be more versatile.

The delay time could be made theoretically as large or as small as desired without any degradation of the signal, and once the signal was in the digital domain it would be easy to carry out other transformations on the signal if so desired.
Once again popular electronic magazines have used this approach [22], but the main disadvantages of the one referenced is that it only uses eight-bit resolution and can only incorporate two fixed delay times.

Pleasing surround sound effects can be generated from some sources and types of music not specifically encoded for this purpose. A large amount of recorded music is subject to signal processing before being mixed down onto the final medium. This can be anything from adding artificial reverberation to the use of Q Sound (section 1.2.2). Each process changes the overall sound and quite often out of phase information is generated on the recording. This can later be taken from the recording and used for the surround sound channels.

It may depend upon the type of music as to which would benefit the most from this effect. Experimentation with different types may yield unexpectedly pleasant results.
Similarly live recordings of music may recreate the ambience of the hall or audience upon reproduction.

The system which I have constructed before and have chosen to do now has its origins in the Hafler (section 1.1.1) and Dolby (section 1.1.4) designs. FIG 1.2 outlines this.

FIG 1.2 Authors Surround Sound Processor Design

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