Audio DSP at NSCA 2001
by Michael Karagosian
©2001 MKPE Consulting LLC All rights reserved worldwide
Published in the June 2001 issue of System Contractor News
Digital signal processing only 10 years ago was not well-known to the architectural audio industry. Not that it hasn't been around awhile. Probably the earliest sophisticated digital processors were the studio digital reverb units that came into being in the late 70's. Outside of the studio console and audio workstation business, however, digital signal processing has not been quick to make its mark on the architectural audio industry./p>
Audio products using DSP have evolved significantly over the years. The earliest products could only run a single program. The signal processing box was still a box, only now with digital words running around inside instead of analog signals. A later generation stored multiple programs that the user could select, making the box more flexible. Some products appeared in the late 80's that could be anything from an equalizer to a delay line. But technology driven development doesn't always grab the marketplace, and these ideas proved to be not so attractive to users.
As the personal computer grew up, so, too, did digital signal processing. As families of microprocessors became popular with programmers, so did families of DSP chips. The combination of available talents became the foundation for a new generation of DSP products.
The simpler of these products is the class I term "fixed DSP". These products are based on algorithms that are complete from input to output, but can be replaced by the user with different algorithms as desired. This style of product has become popular for many speaker processors.
A more generalized class of product, however, are those that I term "variable DSP". These products are based on a family of pre-programmed algorithms, but the algorithms do not define end-to-end signal processing paths. In other words, they do not spell out how the product works from input to output. Users select the algorithms they need, hook them together (usually by means of a GUI -- graphical user interface), and define their own unique signal processing path.
The most popular of the variable DSP products out today is MediaMatrix. Developed by Peak Audio, and introduced in 1993 by Peavey Electronics, this product has become the pièce de résistance of variable DSP design, and has spoiled many a system designer as well. User beware: variable DSP can be addicting.
Besides offering unprecedented flexibility, DSP products also solve a more fundamental problem for audio system designers. Concentrating a large number of DSPs in a centralized location has proven economically attractive. Meaning that one can place all necessary signal processing and routing needs in a single box, rather than in lots of little boxes. Not only is rack wiring simpler, but system control issues disappear.
If we go all the way back to the early 90's, the talk of the day was how to connect lots of little boxes together with networks. The search was on for the ideal network that would transport both audio and control, allowing lots of little signal processing boxes to interconnect and work together. In reality, that ideal network (or bus) exists today in the form of IEEE1394, but as an industry we've changed our focus, and haven't really shown a lot of excitement over this profound development.
Instead, we find ourselves connecting lots of little virtual signal processing boxes on our computer screens. What has cemented this technology into our systems, however, is the flexibility of the GUI. Successful products allow users to create their own system control screens, taking this game to a new level. Now we can create custom GUIs designed specifically for our customers, giving them access to just the features that they need to access. Yet the technology allows us to provide a back door for ourselves, where we can tweak the complexity of our designs, and keep those sensitive settings all to ourselves.
However successful the variable DSP idea has been, the barrier to entry has been high enough to keep most manufacturers standing by and watching. This year's NSCA show, however, demonstrated that the barrier is now significantly lower. A large number of variable DSP products were on the show floor. Previously announced products are now shipping, and newer entries have been announced and will be out shortly. For those contractors and consultants who haven't jumped into the DSP world yet, you're in for a treat. For those who are already addicted to this style of system design, the selection of products and the benefits of competition will be welcome.
Having said this, let's quickly review these cool variable DSP products, complete with flexible GUIs!
Biamp made big news on the NSCA show floor announcing their Audia DSP system. This product is based on 8-channel bricks, uniquely interconnected with Cobranet. The introduction of Cobranet as a product network, rather than just an I/O feature, allows Audia systems to network both audio and control through Ethernet switches to other Audia systems. Audia is expected to ship later this year.
BSS displayed their Soundweb DSP engine. Soundweb also is organized into 8 channel bricks, one rack unit tall, and uses Cat 5 wiring to interconnect boxes. However, BSS does not employ Ethernet on the Cat 5, using instead a proprietary network to carry both digital audio and control. I/O choices are AES/EBU and analog. BSS also makes a clever LCD-equipped remote control panel for Soundweb that fits into a 3-gang box.
IED, one of the early adopters of digital audio signal processing with their UDAPS system, was showing their 3200 Series DSP signal processor. This flexible unit handles 72 audio I/O points per frame, with the split between inputs and outputs determined by the system designer. Input cards can be configured for mic or line levels, and output cards are available as line drivers or as 8W speaker amplifiers.
IRP was previewing their IRP.NET DSP processor. Intended to be a very flexible and expandable processor, each box in this system can handle 8 ins and 8 outs of analog I/O, but can interconnect with other IRP.NET boxes, supporting up to 60 additional audio channels in each direction. Cat 5 cabling or optical fiber is used to interconnect the boxes, but Ethernet is not employed on the Cat 5.
LCS (Level Control Systems) demonstrated their Matrix3 DSP processor with several new features, including Cobranet I/O, their Wild Tracks SCSI-based audio storage, and NetMeters real-time metering capability. Oh, and not to forget their VRAS Variable Room Acoustic System, for artificially changing acoustics, which runs on the Matrix3. (If you haven't heard VRAS in action, you're up for a pleasant surprise.) With their unique selection of BeOs and Linux OS systems, LCS offers great flexibility in networked control.
Peavey was showing the X-Frame88, their latest addition to the MediaMatrix X-Frame line. Adding to the "affordable" tier of MediaMatrix products, this single rack unit box provides the processing power of a full-fledged MediaMatrix single DSP-card frame. As with the older X-Frame, this unit does not require a computer to operate it. Up to 24 ins and 24 outs of analog I/O are available for this device.
Last, but certainly not least, is Yamaha. Yamaha has been previewing their DME32 Digital Mixing Engine for over a year, and began shipping late last year. The DME32 can be operated with or without a computer attached. It is based on a 32 channel bus, supporting analog and AES/EBU I/O. Four DME23 engines can be cascaded to extend the I/O capability to 128 x 128.