The May meeting of the Boston AES chapter was hosted in Cambridge at the new multi-room mastering facilities of M-Works. Founder Jonathan Wyner welcomed members and detailed his recent relocation at 432 Columbia Street and his partnership with DVDLabs, a leading provider of DVD solutions since 1998.

Ira Leonard introduced Roger Talkov of DVDLabs to begin a three-part evening of overview, seminar, and demonstration highlighting the emerging standard, the underlying technology, and the down-and-dirty ins-and-outs of DVD-Audio.

What is DVD-Audio

DVD-Audio is an advanced Audio format, based on the same form factor as CD-Audio, but with vastly enhanced storage, playback and interaction capabilities. DVD-Audio extends the existing family of DVD-Video (set-top playback) and DVD-ROM (PC playback) and DVD-Hybrid (set-top playback with enhanced features for PC) with two more formats: DVD-Audio Only and DVD-Audio/Video.

DVD-Audio Only discs are intended to function just like CD-Audio: played back without a video interface, using track numbers and Play-Stop-Prev-Next functions only. DVD-Audio/Video discs incorporate most of the DVD-Video specification and are intended to take advantage of playback with an integrated video display. Key features include multiple audio streams and multiple supported audio formats, real-time text for onscreen display, full-screen motion video, interactive slide presentations and navigational menuing.

DVD-Audio's most impressive feature is its support of high resolution Packed PCM (PPCM) audio, at resolutions of up to 192kHz 24-bit. As compressed through Meridian's MLP (Meridian Lossless Packing), this non-lossy encoding scheme allows a standard DVD to contain up to 125 minutes of 192k 2.0 source, or 86 minutes of 96k 5.1 source material. Many other permutations exist, of course, such as 258 minutes of 48k 24-bit stereo or even 76 minutes each of 96k 24-bit stereo and a combination of 96k 24-bit L,C,R with 48k 24-bit LR, RR surrounds.

The specification contains built-in functionality to create a stereo downmix from multichannel source material; individual channel coefficients may be set in the software to specify a downmix matrix for playback when the player detects a 2.0 system. DVD-Audio also allows "scalable" audio, or channel-by-channel resolution options, to optimize storage allocation. For example the L,C,R channels can be encoded at 96k 24-bit, and the surrounds can be encoded at 48k 16-bit, allowing the producer to make the best use of the total disc space available.

DVD-Audio also supports popular formats such as DTS, AC3, and MPEG. This compatibility is important to the format's acceptance in a world where the installed base of players ranges from high-end audiophile DVD-Audio only players to first-generation consumer decks, geared primarily to the playback of DVD-Video, to desktop computers with unpredictable playback configurations.

With so many different types of DVDs playing back in a diverse selection of devices, the industry is quickly adopting a "universal" strategy to increase market acceptance and reduce production costs. In order to achieve maximum compatibility with minimum inventory, authoring facilities have been creating "universal" discs designed to deliver the highest-available feature set, regardless of the intended platform.

Each disc is designed with high resolution content for audio-only players. Navigation menus, motion video, and browsable slides are included for audio/video ("universal") players. The same visual content is paired with DTS, AC3, or LPCM content for video players, and special ROM content (weblinks, MP3, standalone applets) can even be included for PC playback.

This comprehensive approach to planning and producing titles is creating a wealth of opportunities for recording artists, music producers, record companies and multimedia content creators. With DVD's phenomenal adoption rate (approximately 6x VCR and CD), the new format looks poised to deliver both artistic and commercial success on an unprecedented scale.

How to Create DVD-Audio

Because the DVD-Audio format includes more complex variables and more disparate sources than either a standard audio title or a DVD-Video project, even greater emphasis must be placed on the pre-production process. The first step is to determine the disc's overall structure: will it contain stereo, multichannel, video, slides? Next the disc's bit budget must be calculated to determine the tradeoff between audio resolution and disc space.

A number of commercially available tools are used to calculate estimates for different combinations. Because each piece of audio will compress differently, variations can and do occur. Even though it's known how classical, pop, jazz, even metal tend to perform during encoding, this estimation is still more an art than a science. Often, this stage includes a complex arbitration between different content and features. For example, how should the final surround mix be executed: 5.1, 6.0, 5.0? Can some channels be encoded at lower resolution? Should the stereo be downmixed by the player on the fly, or does a separate 2.0 mix need to be included?

Also, what type of visual assets will be included: menus, slides, video? Careful planning is necessary to eliminate any surprises during production. For instance, browsable slides, viewable during audio playback, are limited to not more than 2Mbytes of combined file size without stopping playback and reloading the player's video buffer. This means, even with tolerably aggressive slide compression, the number of slides is limited to somewhere from 12-16 per track. However, if the source material contains ambience between tracks, there is no digital black during which to load the next slides and the project is limited to 12-16 slides for the entire program.

Once these parameters are set, final audio and video assets can be prepared. Generally, audio can be delivered to an authoring facility in one of many ways: as an Exabyte, DLT, or AIT--with EDL, on Genex, or on DA98HR. Any separate stereo mixes must correspond exactly with the timecode of their surround couterparts. Full-motion video and accompanying audio should be transmitted on D1 or Digital Betacam, with surround on DA98 as necessary. Slides and menus are created in Photoshop as 720x540 RGB TIFF files. Type smaller than 18 points should be avoided for onscreen display; NTSC-safe colors must be used, and the standard video safe zones must be observed.

After these assets are collected and encoded, the video-player portion of the project is assembled in a DVD creation tool, such as Sonic's DVD Creator or Sonic Scenarist. This partial project must then be merged with a DVD-Audio script to create the master project file. Because there is not yet a commercially viable DVD-Audio tool with a GUI, the creation of this audio script is a difficult and demanding task. Typical scripts include 15,000-30,000 lines of code which must be handwritten or cut-and-pasted from master templates and then modified for the individual project. The script files tend to be large and complicated to perfect, since the specification is not especially fault-tolerant.

Once a completed project master exists, it is necessary to proof and debug. In order to proof the final project, there is currently no alternative but muliplexing a complete disc (software that uses the merged script to combine assets and navigation into disc-ready files) and burning a DVD-R. For full-capacity discs this can often require one hour of multiplex time and one-hour of burn time, per disc. Proofing simulation tools will no doubt come to market in the future, but for now it is a slow and painstaking process to deliver a fully QC'd DVD-Audio disc.

Tour and Demonstration

Following a detailed Question-and-Answer session, M-Works and DVDLabs invited the members to a detailed tour of their facilities and a live demonstration of their latest DVD-Audio titles.

The surround authoring suite, equipped with Genelec surround speakers, and a Sony WEGA monitor, featured Telarc records' "Celebrating Weather Report." This ground-breaking DVD-Audio disc features 96k 24-bit stereo, 48k 24-bit 5.1, as well as MP3, motion video, dozens of browsable slides and an intergrated weblink to www.telarc.com . The surround mixes take aggressive advantage of the format's clarity; the 2-channel mixes reveal the incredible dynamic range.

The flagship surround mastering suite, with Dunlavy SC5 loudspeakers and Spectral poweramps, offered a look at an updated classic--Telarc's landmark recording of Tchaikovsky’s "1812 Overture." With completely updated surround and stereo mixes, including new material recorded exclusively for the DVD-Audio release, this disc includes all 96k 24-bit high resolution audio. The clarity of this format offered surprise after surprise captured by Telarc's KU-100 setup, with the Kiev Children's choir, the massive carrillon at the Church of the Heights, and a masterful outdoor artillery session. By the time the final cannonade finished sounding, there was little doubt about the promise of the medium.

M-Works and DVDLabs concluded the evening with an offer to serve as DVD-Audio informational resources to any Boston AES members interested in creating DVD-Audio titles, by phone 617.577.0014 or web info@dvdlabs.com, free of charge.

- Roger Talkov