UK Conference 1999:
Audio - The Second Century

Church House, part of Westminster Abbey, was the location for 'Audio: the second century' conference in early June.

Papers were presented by society members and industry authorities on a host of subjects relating to audio and its technological and predominantly digital future. Subject matter spanned surround sound to use of home computers in audio. Day two continued the theme of combating threats to audio quality, with especial reference to lossy digital data formats. The afternoon focused on Delivery, that is, delivery formats and hardware. Attention here was on competing high-resolution consumer formats, namely DVD-Audio and SACD (plus Dolby Digital for cinema surround and DTS for multiple surround applications).

Various guidelines are being written for loudspeaker positioning and alignment in surround sound installations, notably Recommended Practice SSF - 02/1-E (3-5-99) from the Surround Sound Forum, for repeatable and reliable multi-channel installations. Gunther Theile's presentation led to discussion of the purpose of the LFE channel (ie, non-essential LF effects, in contrast to the widely-held belief that it is a 'sub-bass' channel for music) . Standardization of channel ordering also raised the issue of metadata, embedded data in a digitstream that provides data about the data.

From the University of Surrey, Russell Mason spoke of an investigation into surround sound mic techniques. In a controlled subjective test, using a cross-section of audio samples (music, reportage, ambient effects), different mic positions were explored, resulting in a general preference for a delayed mic technique, where the rear-facing (but physically close) mics' output was delayed by the temporal equivalent of 10 metres. Of interest was the variation in preferences depending on the source material, eg, spoken voice especially was found to be spatially confused using a large actual separation for the rear microphones.

Holophonics and ambisonics were briefly resurrected by Dave Malham of the University of York, talking of homogeneous and non-homogeneous surround sound systems. High fidelity being the goal, the question was posed: what is the limit imposed by the current paradigm of typically five discrete sound sources spaced around a listener, when a homogeneous sound field is required to seamlessly recreate a sonic event? Since the technologies of today would appear so alien to a visitor from the last century, what could the next century offer in the way of wave field synthesis?

From CEDAR Audio in Cambridge, Christopher Hicks spoke of the choice fading digital audio designers, between general-purpose microprocessors and dedicated DSP chip solutions, a choice between audio precossing by a flexible and powerful microprocessor, versus a rigid yet efficient DSP. A useful paper to put into context the distinction between divergent digital processes.

Echoing this background of theory was a discussion of one-bit digital processing from James Angus of the University of York, The latest challenge facing engineers (PCM processing techniques having evolved to the current state of ubiquity) shows sigma-delta encoded data posing new problems which can be addressed by general purpose DSPs, but preferably an FPGA or custom LSI architecture. These chips are used to construct filters that can process one-bit signals directly.

Television increasingly deems to rely upon the virtual studio, where presenters and actors move around an artifically-created environment. The visual sense may be fooled this way, but any missing (or misleading) natural ambience will betray the production lie. Robert Walker from BBC R&D at Tadworth spoke of an experimental unit to simulate a realistic acoustic as a performer moved around a 'blue screen' set, based on an IBM PC running Windows 3.11, with Lake DSP Huron processing. This audio aspect of virtual production may be in its early days, but its future development must be essential and inevitable.

Digital Mixing Systems were explored by Joji Kuriyama and colleagues of the TOA Corporation, flying in from Japan with desk hardware on show to boot. The presentation constituted a dissection of the company's ix series systems, used for on-air broadcast and live sound control.

To conclude the first day's presentations, the integration of computers and audio was aired. Mark Yonge of SSL explained the problems of audio data file interchange in a post-analogue tape environment. This is the trend for format incompatibilities brought on by competing manufacturers modifying their products for best performance, at the expense of universal interchangability. Jean-Marc Vernier of Digigram explained some of the problems where audio and video digital data must be synchronized. In the professional broadcast environment the challenges have been met, but the knowledge now needs to filter down to home PCs. This is made possible with additional video and audio hardware cards.

Quality of Service is a moot point when data is transferred over a network, like the Internet. Andy Bailey spoke of some of the problems, and failures, besetting the task of high quality audio data transfer over a computer network. With talk of new Internet Audio standards, some of the options under investigation such as Packet Switching, Circuit Switching and IEEE1394 networks were explained. Since good audio has such high bandwidth requirements, technologies like ATM Circuit Switching and FireWire will become increasingly necessary, if QoS is to be established. And within the home PC itself, good quality audio is only now being appreciated. as Steven Harris of Crystal-cirrus Logic explained. Microsoft and Intel, with their combined monopoly of home PCs running Windows software, publish a document every year which sets out minimum requirements for PC performance standards, This now embraces standards of audio quality, tested by inspecting the integrity of analogue and digital audio paths, previously of mediocre quality compared to, for example, Sun and Apple computers. With home PCs becoming increasingly multi-media capable, the goal of improved attention to audio quality here is an important pursuit.

Compressed digital audio data is commonplace, not just in domestic delivery formats but broadcast environments such as ENG and post-production transcoding. The ATLANTIC project was aired, a development of various techniques to combat problems encountered in cascaded bit-rate reduction. The BBC R&D team, which included Andrews Mason and McParland, showed a solution in a ‘mole’ signal that perpetuates through diverse processes to maintain integrity of the original signal. A demonstration of the effects of re-coding on an MPEG layer 2 music sample showed the potentially devastating effects to be avoided — a swimmy electronic babble that had all but obliterated the semblance of music.

For lossless coding, as employed in 1-bit DSD recording, there are other problems to overcome. Simple low-complexity prediction can be applied, as well as more complex adaptive prediction, and a team from Philips Research in Eindhoven presented their experimental predictions. Lossless coding of one-bit audio was shown as a weapon in the combat against threats to audio quality. At the other end of the quality scale, Oliver Kunz from the Fraunhofer Institute stressed the potential of the unfolding MPEG-4 multi-purpose codec. This will be a complete toolbox to take on anything from low bit-rate speech coding up to ‘transparent audio quality’, with data rates from 2 to 300kbits/s. With a variety of Digital Radio systems arriving worldwide, MPEG-4 — derived from MPEG-2 AAC — is a choice for terrestrial and satellite transmission.

Multi-channel audio, touted as the panacea to the industry, requires new investment on everyone’s behalf, from production through broadcast to consumer. In the broadcast arena the infrastructure is set up for two-channel audio, and any changeover necessarily requires expensive reinvestment. Dolby Laboratories has devised a workaround in the form of its new Dolby E codec. Tony Spath of Dolby explained the principles of this system, which allows a discrete multi-channel signal (from 6 to 8 channels) to be lossily compressed to fit through a 2-channel studio’s infrastructure. This includes the use of AES/EBU for multi-channel audio delivery.

Lossy codecs are being used creatively as a possible means for objective evaluation of loudspeaker resolution. Richard Salter of Celtic Audio proposed the use of codecs such as MPEG layers 2 and 3, and Dolby AC-2, to test a speaker’s resolving power — an experimental variable bit-rate codec may allow the revelation of a point at which the increasing codec artefacts are masked by the limits of the speaker’s resolution. Other techniques tried included the deliberate increase of jitter in a CD player, through a modified clock circuit, to again isolate the point at which a given speaker can show faults in the replay chain.

Dolby was doubly represented by John Couling who emphasized Dolby Digital’s role in multi-channel. A demonstration was given of the system’s embedded data (the metadata offering among other things ‘normalization’ of levels) by playing back varied programme material with and without automatic normalization applied.

Francis Rumsey of the University of Surrey gave a résumé of the impending super formats, DVD-Audio and Super Audio CD. Given without prejudice, it served as timely reminder of the potential capabilities of both rival systems, and some of the intrinsic complexities that a consumer may have to overcome to make use of whichever format catches the market.

Chris Hollebone of DTS presented a marketing perspective of his company’s audio format, showing that DTS audio coding, with its variable bit allocation, has the capability to outperform other lossy multi-channel codecs and even certain linear PCM applications.

To round off the conference, freelance technology correspondent Barry Fox voiced his concerns with a speech entitled ‘Is the industry stark raving mad?’ Citing the cases of other high-minded technologies that failed dismally through poor understanding of market needs or usurpment by competing (and incompatible) systems, Barry’s rhetorical presentation was a light-hearted yet thought-provoking ‘outsider’s’ view of the progress of recent developments, particularly with reference to the inevitable DVD/SACD format battle. He finished his warnings with an incitement to all present to find DVD Region Code hacks and spread them far and wide!

Certainly a diverse conference covering plenty of ground, there was nevertheless a core of ideas revolving around the future of audio in the next century, and ways in which to ensure its quality and integrity is not undermined in the scramble for a faster and cheaper world.

Andrew Harrison