Bridging the Gap

Nov 1, 2002 12:00 PM, By Michael Cooper

One can't help but notice the increasing number of 24-bit, 96 kHz-capable digital audio devices flooding the pro-audio market these days. The unprecedented sonic performance that high-resolution digital mixers, recorders, converters, and DAW I/O boxes promise has many owners of project studios and small-commercial studios pondering whether they should upgrade their gear to the emerging standard. But before you whip out your credit card, there are some major issues to consider.

For one thing, just because two pieces of equipment are "24/96" (that is, 24-bit, 96 kHz) capable doesn't mean that they can be used together. They may, in fact, each support a different and therefore incompatible data format, thus preventing them from talking to one another. Many products also demand trade-offs in functionality — for example, a halving of the maximum number of available tracks or channels — in return for working at higher sampling rates. And before you begin production, you'll need to consider which format to use to deliver your high-resolution mixes to the mastering or authoring house.

That, in turn, brings up another question: what end-user/mass-duplication format will your musical masterpiece be released on? If your production is destined for release on CD, will you even hear the sonic benefits of high-resolution recording and mixing once your masters have been downsampled to the CD's 16-bit, 44.1 kHz format? Considering all the hoops you'll need to jump through to produce a 96 kHz recording (not the least of which is monetary), you'll want to be sure that your efforts will result in a superior-sounding end product.

In this article, I'll explore many of the issues that you'll need to resolve in the course of cobbling together a 96 kHz-capable studio and producing a high-resolution master. Most of what I'll discuss applies equally to working at an 88.2 kHz sampling rate, by the way, but for the sake of simplicity I'll refer mostly to 96 kHz digital audio production. (For a discussion about when to use an 88.2 kHz instead of a 96 kHz rate, see the sidebar "Which Sampling Rate Should I Use?")

The touted benefits of working at high sampling rates are clearly controvertible. Indeed, some industry notables contend that the emperor has no clothes. Before rushing headlong into outfitting your studio with expensive 24/96 gear, you should consider the issues carefully. Let's get started with a quick review of digital audio theory.

RATE HIKE

The Nyquist theory states that the highest frequencies that a digital audio converter can capture or reproduce are equal to half the sampling frequency (also known as the sampling rate). In reality, the captured bandwidth is usually a little bit less than half that of the sampling rate, but we're niggling here. For example, a 48 kHz sampling rate should theoretically be able to capture audio frequencies as high as 24 kHz. Thus, a 96 kHz rate should be able to capture twice the bandwidth afforded by a 48 kHz rate, that is, audio frequencies up to 48 kHz.

In order to avoid aliasing tones (unmusical frequencies) from being generated and reproduced by the converter, a steep digital lowpass filter is typically inserted before the converter's output at approximately the Nyquist frequency, or half the sampling rate at the top end of the captured audio band. These steep filters are notorious for generating their own artifacts, such as phase shift and passband ripple (low-level echoes or ringing). Using a higher sampling rate allows you to move the filter higher because the Nyquist frequency also moves higher. With the filter placed far beyond the audible range, a more gradual roll-off can be used, which results in fewer artifacts being generated.

Some proponents of 96 kHz digital audio contend that it's the gentler filter that is responsible for the improvement in sound quality produced at high sampling rates. However, Richard Elen, vice president of Marketing at Apogee Electronics, maintains that this is a ridiculous argument because oversampling at 44.1 or 48 kHz already accomplishes the same thing — that is, it also allows for use of a gentler filter. "These days," Elen explains, "you never actually sample at your real sampling rate because converters do oversampling. Oversampling runs your clock at a multiple of the actual sampling rate. Because your effective sampling rate is so high, half your effective sampling rate is also ultrasonic. That means that all your filters can be way up there, as well. All problems with Nyquist go away. Any ringing or phase-shifting effects caused by the filters will be long gone by the time you get down to the audible range of the audio band. That said, there definitely are improvements that most people hear when listening to high-quality 24/96 conversion systems versus high-quality 24/48 conversion systems."

YOU ANIMAL, YOU

Though many people assume that the improved sonics of 96 kHz digital audio can be attributed primarily to the extended frequency response afforded by the higher sampling rate, Bob Katz, president of Digital Domain (a CD mastering house located in Altamonte Springs, Florida), views the situation a bit differently. "Some people think human beings can hear like dogs and bats," he says, "but in fact we haven't suddenly developed supersonic hearing." While Katz agrees with Elen about the benefits of oversampling, he points out that high-resolution masters destined for CD release will be downsampled to 44.1 kHz rate at some point and thus will have a steep filter and sampling-rate converter applied. "This means there's one more filter in the chain," Katz says, "and it will add its own phase shift, ripple, and other artifacts. You can oversample it all you want [during subsequent playback], but still you've added an additional filter."

Keith Olsen, corporate director of Global Market and Product Development for Mackie, recalls attending an A/B listening test that pitted 96 kHz conversion against 48 kHz, using the same converter. (Readers may recall that Olsen produced many hit records for Fleetwood Mac, Foreigner, Pat Benatar, Rick Springfield, Sammy Hagar, Santana, and others before joining Mackie.) "It sounded like the top end [with 96 kHz conversion] was not as harsh," says Olsen. "But the amazing thing was that upsampling [that is, converting a previously sampled 48 kHz recording up to the 96 kHz sampling rate] sounded just as good! I couldn't tell the difference."

Katz maintains that Olsen's experience proves his contention that the sonic benefits of 96 kHz sampling are due to gentler filtering, and not to extended bandwidth. He points out that upsampling cannot add frequencies that weren't captured in the first place by a lower-resolution A/D converter. "If 50-year-old ears — which typically cannot hear above 15 kHz — can hear the difference between 44.1 kHz sampling and 192 kHz sampling," says Katz, "then obviously what is going on is a reduction in artifacts in the audible band."

In fact, Katz routinely upsamples 44.1 kHz material to 96 kHz when mastering material in his studio for CD release. "When using nonlinear processing (such as compression) at a higher sampling rate," says Katz, "distortion in the audible band is reduced by at least 3 dB, even if followed by a sampling-rate conversion back down to the lower rate." In other words, if you're going to use digital dynamics processing, upsampling the material will reduce the amount of audible distortion in the final file.

Despite his love for the sonics produced by 96 kHz sampling, Katz thinks such a high conversion rate might not be necessary if impeccable digital-filter designs were employed in converters. He also notes that upgrading the digital filters in converters would be considerably less expensive than retooling studios for ever-higher sampling rates.

Elen concurs. "There is a point at which multiplying by large numbers, whether it's high sampling rates or high numbers of oversampling, is done to impress people — in terms of selling things — and not for audio quality," Elen says. "But after a certain point, you're not going to be able to hear the difference anymore."

NOT HEARING IS BELIEVING

Though Elen, Katz, and Olsen differ regarding why 96 kHz converters sound better than lower-resolution ones, they all agree that the difference is audible. But Paul Lehrman, a composer, educator, and consulting editor for Mix magazine, feels that most listeners are unlikely to hear the difference. He contends that "99.999 percent of the people who listen to recordings are not in a position to perceive any difference between a 96 kHz recording and a 44.1 or 48 kHz recording. I think that whatever advantages you get out of 96 kHz are far overshadowed by the limitations of the transducers at both ends of the signal chain."

In defense of his position, Lehrman points out that the frequency responses of most mics and digital musical instruments roll off at around 20 kHz. Thus, anything recorded above 20 kHz at a 96 kHz sampling rate "is probably junk," claims Lehrman. In response to the argument that it's the digital filter in 96 kHz systems, and not the extended frequency response, that's responsible for the improved sonics, Lehrman says that, in A/B tests, he has "never been able to tell, definitively, the difference between a well-constructed 44.1 or 48 kHz oversampling converter and a 96 kHz converter."

Engineer, producer, and EM contributing editor Larry the O weighs in somewhere between 96 kHz cheerleaders and naysayers. "There is some extra sparkle at 96 kHz; it does make a difference," he says. "But it's nothing compared to the difference between 24-bit and 16-bit digital audio."

Naturally, you should listen yourself and draw your own conclusions about whether 96 kHz digital audio sounds better than, say, 48 kHz. Assuming you hear enough of a difference to compel you to take the leap, you'll then need to determine which high-resolution digital audio formats are being used by currently available products and what trade-offs, if any, each format requires.

SPLIT DECISION

The various stereo and multichannel digital audio formats that have been in use since the 1980s were not originally designed to pass 24/96 audio. In some cases, such as with AES/EBU, the spec had enough bandwidth that it could be formally rewritten to accommodate higher sampling rates over existing cables without much ado. In other cases, such as with the 8-channel ADAT Lightpipe and TDIF formats, sample-splitting schemes had to be developed by manufacturers to allow their equipment's multichannel I/O to pass 24/96 digital audio.

Sample splitting is a process that splits, typically, one 96 kHz digital signal into two 48 kHz signals (or one 88.2 kHz signal into two 44.1 kHz signals), thus allowing the split signal to be sent along two channels rather than one. Because sample splitting uses two channels for transmitting each original channel of 24/96 audio, the number of available channels that the ADAT and TDIF I/O can transmit at 96 kHz is cut in half, to four rather than eight channels. If this seems cumbersome, limiting, or confusing, bear in mind that sample splitting is a transitional strategy meant to enable high-resolution digital audio production using entrenched connectivity technologies. Rather than wait until a better solution is developed, we can employ these interim formats to create multichannel 24/96 productions now.

Similar to sample splitting, bit splitting is an encode/decode process that breaks up a 24-bit, low-resolution (44.1 or 48 kHz) digital word into two data streams — one 16-bit, the other 8-bit — and records them onto two tracks of a 16-bit recorder. The two tracks are recombined on playback into one 24-bit stream by a device that decodes the split bit stream. Note, however, that the use of bit-splitting formats is fading as 24-bit devices become the norm.

HELLO, GOOD-BYE

All 24/96 data formats require a chip on the sending and the receiving end of the transmission chain that recognizes what protocol is being used. Without the chip in both pieces of connected gear, the units cannot talk to one another. Before outfitting your studio with 24/96 gear, make sure all the pieces that you're considering buying can speak the same language. If they can't, you'll hear nada.

In many cases, 24/96 operation is available only for some of the supplied I/O on a given piece of equipment. For example, the Mackie MDR24/96 modular hard-disk recorder (M-HDR) can record 96 kHz digital audio by way of its Lightpipe connections, but its AES/EBU connections can handle only 44.1 and 48 kHz sampling rates (at the time of this writing, anyway).

The point is, you can't determine that two pieces of gear will provide inter-operable 24/96 capabilities simply by checking to see if they offer the same type of I/O. The best way to determine 24/96 compatibility is to ascertain that both pieces of gear support the same high-resolution data formats. Unfortunately, a comprehensive summary of supported 24/96 data formats is difficult to come by for most currently available products, and usually requires digging beyond promotional literature, spec sheets, and the like. (For a quick-reference guide to some popular 24/96 products and the high-resolution formats that they support, see the table "Resolution Conflicts.")

In addition to checking compatibility of data formats, make sure any products you plan to use together also support the sampling rate(s) you want to work with and can sync to a common word-clock rate. Some recorders must receive 48 kHz word clock (which is doubled internally) to record 96 kHz digital audio. In such a case, you will need a device like the Swissonic AD96 mk2 4-channel A/D converter (see Fig. 1) that can output half the word-clock rate (44.1 or 48 kHz) to your recorder's word-clock input when working with 88.2 or 96 kHz audio-data sampling frequencies.

Finally, if you are working on a DAW, make sure you have tons of hard-disk storage available: 24/96 data demands a lot of disk space!

Next, let's take a look at the data formats that currently are being offered in 24/96 gear. The following is not meant to be an exhaustive list of 24/96-capable gear; it is offered simply to give you an idea of what's available and what's compatible.

DIVIDE AND CONQUER

S/MUX

S/MUX is a sample-splitting technology developed by Sonorus that splits one channel of 24/96 digital audio into two 24-bit, 48 kHz channels (or splits one channel of 24-bit, 88.2 kHz digital audio into two 24-bit, 44.1 kHz channels) for transmission over ADAT Lightpipe I/O. The new MOTU 2408mk3 DAW I/O box (see Fig. 2), the Swissonic AD96 mk2, and the Apogee AD16 and DA16 converters all support the S/MUX format through Lightpipe I/O. Additionally, the Yamaha 02R96 digital mixer, the Alesis ADAT HD24 M-HDR, and Mackie's HDR24/96, MDR24/96, and SDR24/96 M-HDRs all use an S/MUX-compatible format when transmitting 88.2 or 96 kHz digital audio over their Lightpipe connections. (Yamaha, Alesis, and Mackie's high-resolution Lightpipe formats may, in fact, be identical to S/MUX format; exact specifications were not available.) Let's take a closer look at how some of these products implement their own form of S/MUX to record and play back 24/96 audio.

As one would expect from units using a sample-splitting scheme, the above-mentioned Alesis and Mackie hard-disk recorders all have their maximum track counts cut in half when shuttling high-resolution digital audio over Lightpipe I/O — from a maximum of 24 tracks at 44.1 or 48 kHz to 12 tracks maximum at 88.2 or 96 kHz. The new Yamaha 02R96 mixer (see Fig. 3), however, distinguishes itself as one of the few products on the market that does not lose any channels when operating at 88.2 or 96 kHz sampling rates over Lightpipe I/O. Each recombined 24/96 track (returning from an MHDR, for example) shows up on one 02R96 fader, and the mixer's 56 simultaneous input-channel faders are available no matter what sampling rate you use.

MOTU's 2408mk3 also loses half its maximum number of available channels on each digital bank when using sample-splitting schemes such as S/MUX and 96 kHz TDIF. (I'll discuss TDIF in a moment.) So, for instance, each bank of Lightpipe I/O can transmit only four channels of 24/96 audio in S/MUX mode. Thankfully, though, the 2408mk3 records each 24/96 (or 24/88.2) channel of S/MUX- or TDIF-format audio to one track in Digital Performer (DP). (All MOTU I/O boxes are compatible with Mac and PC DAWs and will also work with other applications besides Digital Performer.) On playback, each DP track becomes just one 96 kHz stream that can be sent out one analog output, in any supported format you wish, on the 2408mk3. That keeps operations simple and user-friendly. Moreover, the 2408mk3 can perform real-time format conversion between its inputs and outputs.

For some readers, it's important to note that the Alesis HD24 can only sync to its Lightpipe inputs when recording digitally at an 88.2 or 96 kHz sampling rate. That is, the HD24 cannot sync to its BNC word-clock input when receiving 88.2 or 96 kHz digital audio over its fiber-optic lines, but must instead sync to the clock embedded in the audio bit stream. Lightpipe has a reputation for being jittery, and finicky engineers may wish to evaluate whether the benefits of recording high-resolution digital audio over Lightpipe justifies the assumed trade-off in jitter performance. Fortunately, Alesis also offers an optional analog I/O board for the HD24 that can record at 88.2 or 96 kHz while synced to the unit's internal clock. All three of Mackie's hard-disk recorders (mentioned above) can sync to their respective word-clock inputs when recording through their Lightpipe I/O at any sampling rate, including 96 kHz.

TDIF

As is the case with S/MUX for lightpipe-equipped devices, recording 96 kHz digital audio over TDIF lines usually requires two channels of storage for every 24/96 channel transmitted. For M-HDRs, that also results in a track count reduced by half. Most mixers that support 24/96 operation over TDIF I/O, such as the Tascam DM-24, also lose access to half their faders when operating in this mode (but, thankfully, each 24/96 channel shows up on only one DM-24 fader).

Things are a bit less cumbersome when using TDIF in DAW land: the MOTU 2408mk3 I/O box records each 24/96 digital audio channel received over its TDIF lines to one track in Digital Performer. Note that although the Apogee AD16, Rosetta 96, and PSX-100 converters are 24/96-capable and include TDIF connections, they cannot handle 24/96 operation over TDIF lines.

Apogee ABS96

A proprietary format developed by Apogee Electronics, Apogee ABS96 combines bit-splitting and sample-splitting techniques to allow you to record two channels of 24/96 digital audio on an 8-channel, 16-bit recorder. The Apogee Rosetta 96 A/D and PSX-100 A/D/A converters both offer ABS96 mode. You'll need the latter unit to decode any tracks that have been encoded in ABS96 format.

2-CHANNEL FORMATS

Double-wire AES

Also known as double-wide AES, double-wire AES mode sends one channel of 88.2 or 96 kHz digital audio down each AES/EBU cable. Thus, two connectors are required for a stereo signal, whereas only one connector is needed to send two channels of 24-bit, 44.1 or 48 kHz audio. Use of the double-wire AES format is not very common anymore, because an AES/EBU spec was formalized for "single-wire/double-speed" (88.2 or 96 kHz) transmission of two channels down one AES/EBU cable. That said, Tascam's DM-24 mixer, DA-98HR 8-track recorder, and MX-2424 SE hard-disk recorder; Benchmark's AD2402-96 A/D converter; Prism Sound's Dream AD-2 A/D converter; and Apogee's Rosetta 96 and PSX-100 converters all support double-wire AES mode in addition to the now-standard single-wire/double-speed AES mode. (The Tascam MX-2424 SE requires an option card to enable double-wire AES operation.)

Single-wire/double-speed AES

Like the regular AES/EBU format that we've always used, the single-wire/double-speed mode sends two channels of 24/96 audio down one connector, but at twice the usual sampling rate. Products that support single-wire/double-speed AES mode include the Yamaha 02R96 and Tascam MX-2424 SE digital mixers; MOTU's 896 and 1296 DAW I/O boxes; Digidesign's 192 Digital I/O, 192 I/O, and 96 I/O boxes; Lucid's AD9624, Swissonic's AD96 mk2, Benchmark's AD2402-96, dB Technologies' AD122-96 MK.II, Sek'd's ADDA 2496 S, Prism Sound's Dream AD-2, and Sonifex's Redbox converters; and Apogee's Rosetta 96, PSX-100, Trak2, Mini-Me, and AD16 (with optional card) converters.

Double-speed S/PDIF

Double-speed S/PDIF is an unbalanced version of the single-wire/double-speed AES format, except that the S/PDIF flavor also uses a different voltage and impedance than AES/EBU. Products that support the double-speed S/PDIF format include the Yamaha 02R96; the Tascam MX2424 SE; the MOTU 2408mk3; Digidesign's 192 Digital I/O, 192 I/O, 96 I/O, and Digi 002; the MAudio Duo USB mic preamp; and the Prism Sound Dream AD-2, the Swissonic AD96 mk2, the Sek'd ADDA 2496 S, the Benchmark AD2402-96, the Sonifex Redbox, and Apogee's Rosetta 96, PSX-100, Trak2, and Mini-Me converters. The Edirol UA5 and UA-700 DAW I/O boxes also both support 96 kHz, but not 88.2 kHz, I/O over S/PDIF lines.

COMPUTER CONNECTIVITY

IEEE 1394

Also known as FireWire, IEEE 1394 is an industry-standard specification developed by the Institute of Electrical and Electronics Engineers (IEEE) for connecting consumer audio and video devices to each other and to computers. However, companies such as MOTU and Digidesign also use IEEE 1394 for connecting their professional I/O boxes to DAWs. The FireWire interface on the Power Macintosh G3 and G4 uses Apple's implementation of IEEE 1394. The original FireWire protocol, IEEE 1394a, provides 100 to 400 Mb per second (Mbps) bandwidth. A new implementation of FireWire, dubbed IEEE 1394b, is just around the corner and will provide up to 3.2 Gb per second bandwidth.

The MOTU 896 and Digidesign Digi 002 both use IEEE 1394a/FireWire for bidirectional 24/96 digital audio connectivity with a computer-based DAW.

USB

USB is another type of bus used to get digital audio (and MIDI) data in and out of a computer. All of the USB-based digital audio devices that I'm aware of use the original USB spec, which provides 12 Mbps bandwidth. By the time you read this, however, devices using the new Hi-Speed USB 2.0 protocol should be out. USB 2.0 provides 480 Mbps throughput, which should dramatically increase track counts as compared with current USB capability.

The M-Audio Duo and Quattro and the Edirol UA-5 and UA-700 are examples of I/O boxes that offer 24/96 audio transmission over USB to and from computers. Due to USB's bandwidth limitations, each of these four products can deliver only two simultaneous tracks at 96 kHz sampling rate, and none of them can record and play back tracks simultaneously. Although the USB-based Apogee Mini-Me can output 24/96 audio by way of its AES/EBU and S/PDIF jacks, the unit's USB port can dish out only 44.1 and 48 kHz rates.

From the above discussion of formats (and referring to the table "Resolution Conflicts"), we can see that the Yamaha 02R96 mixer and Tascam DA-98HR recorder cannot work together in 24/96 mode. Neither the Mackie HDRs nor the MOTU 896 and 1296 I/O boxes can communicate with the Tascam DM-24 mixer in 24/96 operation. And neither the DA-98HR nor the DM-24 can receive 24/96 digital audio from the dB Technologies AD122-96 MK.II, the Lucid AD9624, the Sek'd ADDA 2496 S, the Sonifex Redbox, the Swissonic AD96 mk2, or the Apogee AD16, Trak2, or Mini-Me converters. Clearly, it pays to confirm interoperability of all 24/96 gear that you're interested in before you buy.

Now that you have a handle on the formats currently available for 24/96 digital audio production, let's examine what you need to know about high-resolution delivery formats for your masters.

PREPARING YOUR MASTER

If you decide to dive in to 24/96 digital audio production, you'll need to know what delivery formats mastering houses and authoring facilities can accept. Many DVD-mastering and production facilities prefer that you send them a copy of your mixes on a hard drive. Roger Talkov, president of DVD Labs (a DVD-Audio and DVD-Video production house located in Cambridge, Massachusetts), notes that delivering your project on a hard drive "precludes the need to restore [from the masters you provided] and back up. We also get Retrospect backups on Quantum DLTs." DLT (digital linear tape) can hold up to 80 GB of data on a single cartridge.

DVD Labs can also accept AIFF or WAV files on CD-ROM, DVD-ROM, or Alesis Masterlink discs. For surround-sound projects, which contain multiple tracks for each mix, Talkov emphasizes the importance of making sure that all of your files are saved on your DAW timeline so that they play together in sync.

Lance Clark, multimedia engineer at Gateway Mastering and DVD (in Portland, Maine), notes that most of the masters he receives arrive as 24/96 AIFF or WAV files on a hard drive. He also gets a lot of masters delivered on Tascam HR tape (the format used by Tascam's high-resolution, tape-based MDMs). Clark also notes that Gateway can also accept CD-ROMs, but the company does not receive many CD-Rs in Masterlink format. Although Gateway also masters projects for SACD (a competing high-resolution format to DVD-Audio), Clark observes that such work is currently "slim." He says that the number of projects that Gateway masters for SACD release are perhaps only 1 in every 10 or 12 projects it receives.

Digital Domain's Katz says that he often receives 24/96 and 24/88.2 files, even though he masters for 16-bit, 44.1 kHz CD release. He occasionally receives mixes on hard disks in Pro Tools format, "but 99 out of 100 projects that come to me have been on CD-R." Katz notes that an entire album's worth of 24/96 stereo mixes will typically fit onto three or four CD-Rs. He calls Masterlink a "fantastic delivery format" for this purpose, but he is also comfortable using any CD-ROM containing AIFF or BWF-type WAV interleaved files. (Broadcast WAV Format, or BWF, adds time-stamping to regular WAV files.) Because he has encountered incompatibilities among various DVD writers and readers, Katz does not recommend putting your files on DVD-R at this time.

BIT (OF) RESOLUTION

As this article has made clear, many currently available 24/96-capable products suffer substantial trade-offs in functionality in return for promised higher fidelity. If you've had the chance to audition 24/96 audio gear and you like what you hear, the only question that remains (besides affordability) is whether you can accomplish your goals with it.

As Larry the O says, "with most of the 96 kHz devices in the EM readership's price range, all of your facilities are halved as soon as you go to 96 kHz. In the case of mixers, you no longer have enough channels to do a mix of any level of complexity. Just doing bass and multimiked drums will use up most of your available channels at 96 kHz, leaving no room for guitar, vocals, or whatever. If you're doing a small ensemble like a jazz trio where you're only using one or two mics on the drums, then you might be able to do 96 kHz." Of course, as more products such as the Yamaha 02R96 (which retains its full complement of channel faders in 88.2 and 96 kHz modes) become available, the O's concerns will become moot. Looking even further into the future, TDIF and Lightpipe protocols could easily fade away as more people migrate to DAWs equipped with FireWire.

"Doing bit-splitting techniques and doubling up on things," Elen says, "you end up with a lot of real estate being taken up by connectors. Today you have one little connector [FireWire] that will handle 400 mbps." And with the imminent arrival of 1394b, one can't help but wonder how long it will be before all the digital audio devices in our studio will be connected by FireWire into one large peer-to-peer network.

Make no mistake: our industry is in the midst of a radical transition. Only you can decide when is the right time — if ever — to dive in. If and when you do, hopefully this article will have armed you with the information necessary to make informed decisions in this brave new world of 24/96 digital audio.

---

Michael Cooper is an EM contributing editor and owner of Michael Cooper Recording, located in beautiful Sisters, Oregon.

RESOLUTION CONFLICTS

This table shows a sampling of currently available high-resolution digital audio products and the 24/96 data formats they support (indicated by a - mark). As long as they offer compatible sampling frequencies and word-clock rates, any two products that support the same format should be able to work together in 24/96 operation (with the possible exception of products that sport FireWire or USB ports, which are included here only as a handy reference to show which DAW I/O boxes support 24/96 computer connectivity).

The I/O that a product provides is implicit in its supported formats. For example, a product that supports S/MUX format will provide ADAT Lightpipe I/O. Some products may require an option card to enable a particular format that is noted here as being supported.

RESOLUTION CONFLICTS

This table shows a sampling of currently available high-resolution digital audio products and the 24/96 data formats they support (indicated by a - mark). As long as they offer compatible sampling frequencies and word-clock rates, any two products that support the same format should be able to work together in 24/96 operation (with the possible exception of products that sport FireWire or USB ports, which are included here only as a handy reference to show which DAW I/O boxes support 24/96 computer connectivity).

The I/O that a product provides is implicit in its supported formats. For example, a product that supports S/MUX format will provide ADAT Lightpipe I/O. Some products may require an option card to enable a particular format that is noted here as being supported.

	S/MUX or compatible format	96- kHz TDIF	Apogee ABS96	Double-wire AES	Single-wire/ Double-speed AES	Double-speed S/PDIF	Fire Wire	USB
Alesis ADAT HD24	¯
Apogee AD16	¯				¯
Apogee DA16	¯
Apogee Mini-Me					¯	¯
Apogee PSX-100			¯	¯	¯	¯
Apogee Rosetta 96			¯	¯	¯	¯
Apogee Trak2					¯	¯
Benchmark AD2402-96				¯	¯	¯
dB Technologies AD122-96 MK.II					¯
Digidesign 192 Digital I/O					¯	¯
Digidesign 192 I/O					¯	¯
Digidesign 96 I/O					¯	¯
Digidesign Digi 002						¯	¯
Edirol UA-5						¯		¯
Edirol UA-700						¯		¯
Lucid AD9624					¯			¯
M-Audio Duo						¯		¯
M-Audio Quattro								¯
Mackie HDR24/96	¯
Mackie MDR24/96	¯
Mackie SDR24/96	¯
MOTU 1296					¯
MOTU 2408mk3	¯	¯				¯
MOTU 896					¯		¯
Prism Sound Dream AD-2				¯	¯	¯
Sek'd ADDA 2496 S					¯	¯
Sonifex Redbox					¯	¯
Swissonic AD96 mk2	¯				¯	¯
Tascam DA-98HR				¯	¯
Tascam DM-24		¯		¯	¯
Tascam MX-2424 SE				¯	¯	¯
Yamaha 02R96	¯				¯	¯

MANUFACTURERS

ALESIS
tel. (401) 295-9000
e-mail info@alesis.com
Web www.alesis.com

Apogee Electronics Corporation
tel. (310) 915-1000
e-mail info@apogeedigital.com
Web www.apogeedigital.com

Benchmark Media Systems, Inc./Sonic Sense, Inc. (distributor)
tel. (800) 262-4675 or (315) 437-6300
e-mail sales@benchmarkmedia.com
Web www.benchmarkmedia.com

dB Technologies/Lavry Engineering, Inc. (distributor)
tel. (206) 381-5891
e-mail jeff@lavryengineering.com
Web www.dbtechno.com

Digidesign
tel. (800) 333-2137 or (650) 731-6300
e-mail prodinfo@digidesign.com
Web www.digidesign.com

Edirol Corporation North America/Roland Corporation U.S. (distributor)
tel. (323) 890-3700
e-mail edirol@edirol.com
Web www.edirol.com

Lucid
tel. (425) 742-1518
e-mail info@lucidaudio.com
Web www.lucidaudio.com

M-Audio
tel. (626) 445-2842 or (800) 969-6434
e-mail info@midiman.net
Web www.m-audio.com

Mackie Designs
tel. (800) 258-6883 or (425) 487-4333
e-mail productinfo@mackie.com
Web www.mackie.com

Mark of the Unicorn, Inc. (MOTU)
tel. (617) 576-2760
e-mail info@motu.com
Web www.motu.com

Prism Sound/Prism Media Products Ltd. (distributor)
tel. (973) 983-9577
e-mail sales@prismsound.com
Web www.prismsound.com

Sek'd/plus24 (distributor)
tel. (800) 330-7753 or (323) 845-1171
e-mail info@sekd.com
Web www.sekd.com

Sonifex Ltd./Independent Audio (distributor)
tel. (207) 773-2424
e-mail sales@sonifex.co.uk
Web www.sonifex.co.uk

Swissonic/plus24 (distributor)
tel. (800) 330-7753 or (323) 845-1171
e-mail info@swissonic.com
Web www.swissonic.com

Tascam
tel. (323) 726-0303
Web www.tascam.com

Yamaha Corporation of America
tel. (714) 522-9011
e-mail info@yamaha.com
Web www.yamaha.com

---

WHICH SAMPLING RATE SHOULD I USE?

Most people who currently work with high-resolution digital audio prefer using the 96 kHz sampling rate and 24-bit word lengths for their productions. Roger Talkov of DVD Labs cites this as the most common format for both stereo and surround-sound DVD-A (DVD-Audio) discs. That's mostly because 96 kHz is a little easier to work with if your project will also include video content. Read on to learn why this is so.

Despite all the concerns a couple of years ago about whether DVD-A discs would play in early-model DVD-Video players, the DVD disc has turned out to be fairly universal. (Of course, a DVD disc won't play in a CD player.) DVD discs can play in either DVD-Audio or DVD-Video players because audio content is stored in a different format compatible with each type of player and on separate portions of the disc. Many DVD-Video discs use the Dolby AC-3 compression format for encoding their audio content, and the Dolby digital encoder wants to "see" a 48 kHz sampling rate to do this work. Converting a 96 kHz sampling rate to AC-3 is a little easier than converting a rate of 88.2 kHz because 96 is an integer multiple of 48 (2 × 48 = 96). That simpler processor "math" should theoretically result in a more pristine conversion. That's why you're better off using a 96 kHz sampling rate if your DVD project will include video content.

Although the DVD-A spec also includes support for 176.4 and 192 kHz sampling rates, and all DVD-Audio players are capable of playing audio at those rates, they have very rarely been used to date.

In some cases, a studio might be planning to release a project on both DVD and CD. If only one mix will be produced for both release formats, the engineer might choose to mix at the 88.2 kHz sampling rate. Because 88.2 kHz is an integer multiple of the CD-format's 44.1 kHz rate, downsampling an 88.2 kHz mix for CD release should theoretically cause less degradation than downsampling a non-integer-multiple 96 kHz mix. Accordingly, Talkov sometimes receives 88.2 kHz masters for DVD authoring. Gateway Mastering and DVD, on the other hand, reports that it rarely receives 88.2 kHz DVD-A files to accommodate conversion to Red Book CD format. "In fact," Gateway's Lance Clark says, "I think we've done only one DVD-A disk at 88.2 kHz."

Downsampling theory notwithstanding, Talkov says that "there are excellent sampling-rate converters that will take 96 kHz to 44.1 kHz okay." Bob Katz of Digital Domain, however, takes a stronger stance on this subject. He asserts that, considering the pristine performance of the best sampling-rate converters available today, there is no practical reason ever to use the lower 88.2 kHz sampling rate in lieu of 96 kHz. "We're talking one flea's worth of audible difference" between 88.2 and 96 kHz files, Katz says. "Still, I've been pushing for 96 kHz. It seems to be a touch warmer, but only because the filter is just a little bit farther out." Preferences aside, Katz also regularly accepts 88.2 kHz files for CD mastering.