All the publicity in the Internet digital audio market thus far has been centered on the single-purpose MP3 player. While this vertical market application is creating a loyal following and a flurry of new consumer products, some of the hardware designs may be too specific to a single function: MP3 playback. Many of these device designs will not be compatible with all of the emerging Internet audio standards, which will only serve to delay market acceptance of Internet digital audio.
If, however, the decode engine is a general-purpose embedded processor, the system design and hardware technology required to decode digital music will not only be flexible, but surprisingly similar to most vertical PDA devices already being manufactured. It is therefore the specific software solutions that will differentiate the products in the emerging e-appliance market, if the designer allows for these software differentiators in the initial processor selection process.
What will it take to move the MP3 decoding technology into other handheld consumer devices, such as e-books, general-purpose PDAs and information appliances? After all, why shouldn't consumers be able to listen to music while reading a favorite book, or surfing Web sites, without a PC or multiple devices? Forecasts for the MP3 player market vary wildly, from predictions of sales of just over 1 million in 1999 to sales of 5 million or 6 million over the next several years.
While time will tell, it may remind some of us of the Internet TV appliance market forecasts of a few years ago. In the case of Internet TV appliances, the "killer app" for the Internet was and is e-mail and the real discovery for marketers was that no one really wants to read and send e-mail from the TV set in the family room. So while demand for e-mail usage remains strong, the convergence of e-mail access with consumer technology did not evolve as predicted.
In the same way, one could argue that the Internet music player does not really represent a market, but rather a technology that will eventually move into other consumer products. In this case, the killer app is CD-quality music at 10 times the compression ratio and delivered in real-time over the Internet. E-commerce is definitely here to stay and how e-commerce markets Internet music has yet to be determined. Neither has the means through which consumers choose to listen to Internet audio. However the market evolves will have a profound impact on Internet audio playback devices.
And no matter which way the Internet music market evolves, hardware designs will evolve to accommodate new standards and meet market demand. Since the Internet music market is still in its infancy, the hardware technology solution designed for today's digital music players relies on a multichip architecture with a DSP at its heart. The typical design uses an 8-bit MCU to control the system and I/O. Additional glue logic is also required and the system essentially becomes a fixed-purpose decoding engine. One of the challenges of this kind of multiprocessing solution is that it requires two separate software coding and debugging efforts. This makes this kind of design a software designer's worst nightmare.
Another design approach would be to use a general-purpose embedded processor and use specific algorithms to handle the digital music decoding function, all on a single chip. Although this is not a hardware-optimized solution, it does leave open the possibility of adding other functionality to the device. Or, if the electronic appliance were conceived for a different purpose, such as to store an electronic book or phone numbers, adding Internet audio capability later would always be a possibility if the right processor were initially selected for the design.
If the hardware were capable and provided enough upward mobility to add features through software, as with an embedded processor like an ARM 32-bit RISC architecture, it then would become a software personality design approach. An ARM 720T core, for example, will run in excess of 70 MHz and has the software available to make it do just about anything. The MP3 decoding algorithm uses approximately 45 Mips, about two-thirds of the available bandwidth, and the software can be licensed from several sources.
New Internet audio standards from Microsoft, Dolby, Real Audio and Lucent/AT&T are beginning to emerge, making it difficult for a designer to commit to a single fixed-function DSP solution. At the same time, the Secure Digital Music Initiative (SDMI) is trying to get a handle on the security and standards issues, which could have a profound impact on design decisions. The bottom line is that flexibility in hardware design is essential. That's why hardware design choices that allow for upgrades are becoming more popular as well as "safe." Software can be field upgradable, but changing the CPU is a little more involved.
Picking the right chip platform is critical for effective design. Consider, for example, the processing requirements for a typical handheld MP3 device. If you select a microprocessor with a 74-MHz operating frequency to run the MP3 decode software and a total system dissipation near 85 mW, the CPU has plenty of processor overhead still available to handle other tasks, such as your electronic address book. You might also want to think about choosing a technology that provides a direct interface to peripherals, such as an audio digital-to-analog converter. For example, most vertical PDAs include plenty of DRAM and flash memory, so by adding a few extra components a PDA can be modified to include support for Internet audio decoding with only a modest incremental cost .
Time-to-market demands for consumer products are making it even more difficult to accommodate the lead times for ASIC design prototypes. To maintain design versatility and shorten time-to-market, system designers are coming to rely on off-the-shelf standard products with well-supported software tools and development boards as the de facto standard.
Meanwhile, the same rules apply to designing other devices to accommodate digital music. What if you could read an electronic book while listening to CD-quality music? Design variations are certainly a possibility if the hardware design is flexible enough to handle any of the emerging audio decoding software standards. The common thread is the delivery medium-the Internet.
Another application is the ever-elusive Internet appliance. Talk of convergence products that bring all forms of entertainment into one home appliance or set-top box could easily add the digital music player capability. This would not only serve to enable this market, but allow the consumer any form of digital audio technology.
And, what about audio recording or ripping? Encoding capability in the handheld device-without the PC-is a technology that could easily be added with more Mips. ARM 9 products on the horizon could add the extra horsepower needed for tethered as well as portable applications. In fact, the home appliance would be the ideal place for introducing a digital video capability or the harddrive music player.
On the software side, announcements of changes in real-time operating systems are adding new support functions to these emerging markets. Application layers are becoming more complicated and may require a separate RTOS. Having the right processor with MMU and cache is all part of the primary selection criteria.