You are here

Coding Computer Audio

Cutting Edge By Dave Shapton
Published January 2000

Emblaze is a currently available system in which audio data contains the means to decode itself.Emblaze is a currently available system in which audio data contains the means to decode itself.

Dave Shapton keeps track of the latest developments in the coding of computer audio — and the hardware used to amplify it.

The power of consumer soundcards for home computers is growing at an incredible rate. The problem is that developments in computer audio have, until now, outpaced the physical capabilities of the amplifier and speaker systems sold with most computers: so‑called 'multimedia' loudspeakers are usually abysmal. British company Videologic (who, among other things, are a major developer of consumer soundcards) aim to correct this with their new Sirocco range, which represents an interesting application of pro audio design and construction techniques to the consumer market for desktop speakers. The idea of constructing a 'multimedia' speaker system around a very small pair of 'satellite' speakers and a separate subwoofer is not new: Bose produced a similar system about five years ago. However, Videologic's system sounds far better than any other desktop system I've heard. The satellite speakers are bi‑amped and are trapezoidal to avoid internal standing waves, while the powered subwoofer crosses over at about 120Hz, as opposed to the 300Hz that is typical of other multimedia speakers and which falls right in the middle of some important vocal frequencies. At less than £240 for the whole system, the Sirocco stands a realistic chance of success in the consumer market, being ideal for playing MP3s and CDs while you are at your computer.

The standard Siroccos, however, are still not quite good enough for demanding applications like mixing and mastering. And yet these are activities that you are as likely to do these days at a computer as in a dedicated studio. Interestingly, then, Videologic have produced a 'Pro' version of the Siroccos using high‑definition aerogel (HDA) cones. The amp's power has been upped to 100W RMS, they've fine‑tuned the crossovers and added a 24‑bit/96kHz D‑A converter. The whole system, including sub‑woofer, amplifier and DAC, costs £549 including VAT.

To my mind, it's very interesting to see companies like Videologic breaking down the gap between ' multimedia' and 'professional' products. Whether this is a sign that consumers are expecting higher quality or simply an acknowledgement of the depth of the market for affordable, professional‑quality gear, it's a development that can only benefit SOS readers.

Speaking Of Code

You may remember that last month I was getting very excited about the prospect of being able to send the means to decode an audio or video format in the same data stream as the media itself. It has occurred to me that this is such an abstract idea that I really ought to give it some more explanation.

I suppose what's behind all this is the frenzy surrounding MP3. My view on this is that MP3 is the beginning of a journey, rather than the end. Yes, it's a high‑ (enough) quality solution to downloading audio over low‑bandwidth Internet connections, and it's great for playback from portable solid‑state players. But no, it's not the ultimate technology by any means; in fact, it's just one of many possible solutions. My feeling is that in a very short time no‑one will care what format we use, whether it's AC3, MP3 or even uncompressed digital audio — or something else no‑one has heard of yet.

How, then, can we avoid a proliferation of 'standards' and mutually incompatible media devices to play them on? Quite simply in fact: by having players that contain enough processing power to decode any format in software. Then, if your data stream contains information about how to decode itself, any device can decode it. The moment you can do this, you won't even care what format is in use, as long as it sounds OK.

If your reaction to this is along the lines of "yeah, right, but it'll never happen", then have a look at This is an audio and video streaming technology that uses a Java player, sent with the media data, to decode and play the media stream. It works now, on your computer. I don't know what compression format it uses and I don't need to know. What's more, they could change the format every week and I'd be none the wiser — and that's precisely the point. If the means of decoding the audio is sent as part of the audio itself, it can remain totally transparent to the user.

Millennial Angst

With all the attention being paid to the Millennium Bug, you may be surprised to know that software booby‑traps of a similar nature can occur for other reasons. If you have a digital audio workstation, try creating a project with one audio event at time zero on the timeline, and another at around 13 and a half hours. I know of at least two systems that won't let you do this, and I suspect that there are several more. If you do a few sums to work out the biggest number of samples you can represent with a 32‑bit binary number, what do you get? The number of samples in 13.5 hours of 44.1kHz stereo audio! Does it matter? Well, yes, it does. SMPTE timecode is based on a 24‑hour cycle, and if you are working in a studio — especially one that handles video as well as audio — you could start your project at any point in the 24 hours. Start mastering your album at 13 hours and you'll be in trouble halfway through.

The practical fix for this is to put in a timecode offset, if your system will allow it. This type of bug is more serious than one that is the result of a programmer's typo. It's architectural. It's part of the basic structure of the software. You can't change it without a complete rewrite. And more often than not, changes at this level introduce more bugs than they fix. No wonder I still prefer to use a DOS‑based sequencer!