Mass Market Multimedia
Multimedia -- the simultaneous use of multiple, interrelated, electronic media -- is one of the most intriguing aspects of today's personal computer industry. From full-featured computer games and highly interactive educational courseware to sophisticated business presentations and on-line video conferencing systems, multimedia is redefining the concept of computerized information to include completely new data types such as audio and video.
Until now, the journey to multimedia's mass market acceptance has been evolving slowly, hampered by a lack of standards, software support, computer processing horsepower and applications. And although the technology required to implement multimedia has been available for some time, until now it has been prohibitively expensive for all but the most high-value applications.
Today, the multimedia market is poised for tremendous growth as these key enabling factors fall into place. One of the last remaining challenges facing the multimedia market, however, is developing hardware that combines low cost with acceptable quality and performance levels. Once the cost of making a PC multimedia-capable can be brought down from thousands of dollars to hundreds, true desktop multimedia can become widespread.
Cirrus Logic is taking a leadership position in the multimedia market, applying "smart" integration to help create complete low-cost solutions for implementing desktop multimedia systems. Cirrus Logic knows the multimedia market, its dynamics and the technologies required. It is aware of the issues facing the industry, and is intimately familiar with all the technological challenges confronting designers of these systems.
Most importantly, Cirrus Logic has access to the technologies required to create highly integrated multimedia solutions, as well as the uniquely broad perspective -- the total systems view -- necessary to intelligently architect these solutions.
A Blend of Technologies
"Multimedia" is perhaps one of the most overused terms in the computer industry today. As a result, its meaning has become somewhat diffused. In practice, multimedia is not one thing, but a blend of core and supporting technologies working together to serve a diverse range of applications.Multimedia, at its core technology level, encompasses graphics processing, video processing and audio processing, along with supporting technologies such as data compression and decompression schemes, mixing and matching two or more of these technologies as needed to perform the requirements of the application. While it can be said that an application that presents a diagram on the screen and beeps is providing multimedia, the common usage of the term implies a much more sophisticated level of interaction.
Together with this graphics, video and audio processing, a multimedia system usually incorporates important supporting technologies. A number of video and audio compression technologies -- JPEG, MPEG, Px64, Cinepak and ADPCM, to name a few -- enhance the storage and communications aspects of multimedia by reducing the data volume. Some form of mass storage is required, either magnetic or optical, including CD-ROM and large capacity hard disk drives, both for storing and playing back video and audio data. Lastly, local and wide-area high-bandwidth communications are a necessary component of many multimedia systems such as video conferencing applications, and may include FDDI, Ethernet or Fibre Channel networks, as well as ISDN, Switched-56, frame relay and ATM wide-area services
Computer users have come to expect a multimedia machine to be able to handle at least one and preferably many of the following activities:
The graphical components of multimedia include text in a variety of fonts and styles, traditional graphics such as charts and illustrations, photographic still images and animation. A multimedia system may need to capture video input from analog and digital sources, edit and transform these inputs and play them back in a windowed display or save the output to disk. In a similar manner, a multimedia system may need to capture, process and play back audio signals from several sources.
- Accept some form of large, prerecorded mass-storage medium, such as CD-ROM.
- Retrieve from storage and display digitized images, still or moving with text in multiple windows.
- Play back digitized high-fidelity audio, possibly through multiple channels.
- Sample, digitize and store audio.
- Control an external MIDI device.
- Accept analog video from a camera, VCR or videodisc play for frame capture or digital recording through a compression algorithm.
- Provide output in analog video format for recording or distribution.
Multimedia Opportunities
Digital video and digital audio technologies will have a major impact on the way people use personal computers. Already, existing applications are being enhanced by these technologies, and new classes of applications, not possible before, are emerging.Each technology, alone and in conjunction with the others, addresses a number of potential markets. The market for digital video comprises live video-in-a-window, video playback, video editing and video conferencing. The digital audio market comprises games, multimedia personal computer (MPC) audio, business audio, workstation audio and video conferencing.
Video-in-a-window takes video from an analog source, such as a VCR, laser disk or tuner, and overlays it onto the PC's graphics screen through either digital or analog means. It is currently the largest digital video market, useful for training systems, security systems and traditional television viewing. It is relatively low cost to implement, precluding the need for any sophisticated hardware or mass storage.
The video playback market, though currently small, has a tremendous potential for growth. Video playback adds a high degree of interactivity, enabling the user to select video clips and movies on demand and view them on the graphics screen. The most common form of video playback uses CD-ROM for the delivery of games, digital movies, entertainment and educational materials. As LAN technologies develop that support the bandwidth demands of digital video, they will be used more for a variety of video playback applications, such as video mail, presentations and training. Video playback will also find uses in portable presentations played back on a portable computer using hard disk storage. Because of the mass market potential for this technology, cost is a primary concern.
One of the enabling factors for the video playback market is the video editing market, which includes the creation, editing and publishing of video playback materials. Currently, video editing systems are quite expensive, but as costs drop new markets will emerge, expanding over time to a potentially large combined size, just as the appearance of low-cost desktop publishing systems created new markets. A new generation of highly integrated video processing ICs will have a significant effect on reducing system costs.
The largest vertical market today, and one with tremendous growth potential, is desktop video conferencing. Much more than simply a video telephone, video conferencing enables collaborative computing, on a campus or around the world. Engineers can view the same rotating 3-D image of a model, for example, each able to stop, speed up, highlight or add to the image while simultaneously viewing the other participating parties. Video conferencing can also create virtual offices for telecommuters, providing the sense that remote workers are part of the central office. The availability of inexpensive high-bandwidth communications, both local and wide area, is one gating factor limiting the expansion of this market. Multistream video processing ICs and hardware compression/decompression are another gating factor that must be resolved before video conferencing becomes widespread.
One of the oldest multimedia market segments is the games audio market, providing the ability to record and play back sound as well as deliver synthesized music. Given a tremendous boost by the appearance of the Creative Labs SoundBlaster add-in board for PCs in 1990, the market is now shifting to the cost-effective16-bit technology introduced by Crystal Semiconductor. Formerly comprised 100 percent of add-in cards, the market is rapidly moving toward audio subsystems integrated on the motherboard, and has shown strong growth in 1992, with more than 3 million add-in cards sold. The existing de facto interface standard (AdLib for music synthesis and SoundBlaster for audio) is shifting to the Windows interface.
Established by the MPC (Multimedia Personal Computer) Marketing Council, the MPC audio standard requires an 8- or 16-bit PCM (pulse code modulation) codec, CD-ROM, music synthesizer and a mixer with volume control. MPC is supported by the Windows 3.1 Media Control Interface. Growth of this market has been hampered by the slower than anticipated drop in the price of CD-ROM drives. Today, the silicon cost to implement MPC on an add-in card runs about $85, but higher degrees of integration can cut these costs significantly while also enabling these functions to migrate to the motherboard.
Business is beginning to appreciate the value added by the application of audio technology to traditional business communications. Voice annotation of documents and spreadsheets, voice and sound added to presentations, voice recognition and text-to-speech systems are all being experimented with. Although there are yet no formal specifications for this emerging market, Compaq and Microsoft have already entered this market, employing a 16-bit codec for low-cost recording and playback (music synthesizer optional) that is well suited for motherboards and Microsoft's Business Audio Solution software. This is beginning to establish a de facto industry standard for business audio.
In a similar manner, audio is enhancing the use of engineering workstations by providing voice annotation, voice mail and video conferencing. Initiated by Silicon Graphics and Sun Microsystems, all new workstations (after January, 1993) have 16-bit audio systems built in. Both the Unix operating system and Windows NT support audio communications.
Audio is also a major component of video conferencing. Dominated by Compression Labs and PictureTel, the system cost over the past seven years has dropped from $250,000 to $50,000. Using 16-bit audio as the standard, desktop systems are being developed this year that could cost as little at $5000. Recently, Picturetel introduced a PC-based system incorporating Pixel Semiconductor's CL-PX2070/2080 digital video processor and MediaDACTM multi-source video digital-to-analog converter.
Implementing any of these multimedia systems, however, poses a broad range of challenges, including the problem of computational power. Graphics and computer imaging -- especially when combined with video and audio playback and user interface processing -- require large amounts of computational power. Even powerful CPUs have often proved inadequate for the task, and the few machines that can handle the processing load are prohibitively expensive.
Technology Issues
Today, several of the key drivers pushing the growth of the multimedia market are already in place. Multimedia standards in the areas of digital video and audio communications, file formats, compression, etc. are to a large extent established. Operating systems and environments, such as Microsoft Windows, with full support for multimedia applications are here. Many, if not most, PCs sold today have enough horsepower to effectively handle the increasing bandwidth and processing demands of multimedia, and sales of CD-ROM drives -- the principal delivery medium for multimedia applications -- are growing rapidly, with nearly 6 million units shipping this year, tripling to nearly 18 million units by 1996 according to a 1993 Disk-Trend report. Finally, as a result of these standards and supporting environments, as well as the large and growing installed base of capable platforms, multimedia applications are beginning to flood the market.Microsoft WindowsTM will have a major impact on the multimedia market. With the addition of this software, millions of PCs are immediately capable of playing video on the current display without any additional hardware, creating a tremendous impetus for application development. However, the performance of the user's computer system as well as the performance of the application will directly affect the end user's experience and satisfaction level with these applications.
Software playback performance can be broken down into three components: image quality, available bandwidth and usability. Image quality depends on the number of frames per second, the resolution and color depth. The more of each of these, the better the image quality, but the amount of data needing to be processed increases dramatically. For example, full-screen, true color, full-motion video requires 1280 x 1024 x 24 x 30 = 943,718,400 bits of information to be processesed each second. And so the available bandwidth -- the amount of CPU horsepower left over for tasks other than processing the video data -- must be considered. Finally, playback performance depends on the level of usability achieved -- on how easy it is to capture, create, transform and play back video images.
Thus, the capabilities of the computer system -- the CPU, clock speed, mass storage system and display system -- will have a very strong effect on overall software video playback performance and user satisfaction level. Obviously, any hardware than can accelerate key video playback functions can have a dramatic effect on both of these.
Engineers creating multimedia hardware face a number of design challenges. Many multimedia applications, such as video editing and conferencing, require the management of multiple data streams. These applications generally require the multiple, independent, live video windows to be displayed, and some require special video effects such as transitions, fades and advanced chroma keying.
Perhaps the most important function that can improve video playboack performance is high-quality zooming with interpolation and scaling with filtering. The normal video frame size is only 160 by 120 pixels, just 2.5 percent of the total frame area of a display system with 1024 by 768 resolution. The ability to zoom up the images to larger physical sizes without blockiness or image distortion is a key benefit, but increasing frame size increases the data to be processed exponentially. Because the number of pixels in the original frame is fixed, scaling the frame up would have the effect of spreading the fixed number of pixels over a larger area. An intelligent algorithm must be used to synthesize intermediate pixels to preserve image quality, but the processing overhead of such an operation makes it impractical for anything but static displays. Similarly, when reducing the display area, the number of pixels must be reduced without affecting image quality. Ideally, the image is filtered, generating an entirely new set of pixels, with the color of each new pixel calculated using the values of the original pixels. Specialized hardware can perform this task at high speed for video displays without burdening the CPU.
Other features, such as frame smoothing, color space conversion and image enhancement also improve perceived quality. Obviously, special scaling and processing hardware can significantly improve video quality and overall system performance, even for high-end machines.
To complicate matters further, there are a multitude of video data stream formats and data rates, creating significant interfacing challenges. NTSC and PAL, the predominant video standards, have resolutions, frame rates and color formats that are fundamentally different from those of digital video or the standard PC output, super VGA.
Audio sound quality has become an important factor in designing multimedia systems because of the proliferation of CD-quality audio. Good sound quality is a must whenever digital audio is used for presentations, as well as other important applications such as voice recognition and text-to-speech systems.
To achieve this level of quality, the audio subsystem must be able to capture 16-bit audio samples, providing a theoretical dynamic range of 96 dB. The system must be able to capture these samples at a rate of 44 kHz to enable the reproduction of the full human hearing range of 20 Hz to 20 kHz, and stereo operation has to be available. Analog-to-digital conversion technologies such as delta-sigma also affect audio quality by reducing quantization noise and distortion, and audio mixing and volume control technology can eliminate pops and clicks. Lastly, the actual implementation of the audio subsystem can affect overall quality.
Sound quality, however, is not the only requirement of a competitive sound system. Performance is important, too. Audio processing can place a significant burden on a PC. Recording 16-bit stereo at 44 kHz generates 177 kilobytes per second, or more than 10 megabytes per minute that need to be processed and stored.
MPC specifications dictate that audio processing not require more than 15 percent of the available system bandwidth, but this can be easily exceeded on many of today's PCs, resulting in annoying audio distortions and degraded system performance. The challenge, then is to reduce the data rate without compromising sound quality, and to do it without using costly digital signal processors.
Smart Integration
One of the most fundamental ways to reduce hardware costs, from a semiconductor vendor's point of view, is to produce more highly integrated circuitry in the smartest possible way. But unlike other more narrowly focused markets, only a company that has assembled all of the fundamental multimedia technologies can properly combine them to create intelligently integrated, better defined products.For example, there's a fair amount of overlap between graphics and video systems in the PC environment. Both need high-bandwidth interfaces and data paths, both need frame buffers. Live digital video or video playback from a CD-ROM or hard disk are most likely to share the same display device with graphics data, usually in a video window or as a background. If a company only knows how to design graphics hardware, it will have a difficult time learning to integrate video functionality, and vice versa.
Today, graphics functions and video functions are separated and there is no standardized way to bring them together. As a result, solutions are often cumbersome, of low quality and very expensive because the level of integration is very low and there is a lot of duplication of components. If graphics and video data could, for example, share the same interface, frame buffer and display, achieved through smart integration of graphics and video functions, very inexpensive solutions could be created.
In a similar manner, integration of graphics and audio functions could bring cost down significantly while increasing quality and performance. High-performance audio requires a high-speed mechanism to bring digitized sound data from the hard disk to the audio device. Today, audio is managed through direct memory access (DMA), a relatively slow process for a PC. Graphics systems, however, are already using a high-speed CPU interface through VESA and PCI local buses. If this high-speed interface could also be used to transport audio data, and the graphics frame buffer used also as an audio data buffer, then audio management becomes much less burdensome to the system, improving sound quality and overall system performance.
In today's highly competitive market, simply supplying hardware alone is not sufficient for success. Semiconductor manufacturers must be able to support their core technologies with the approriate software conponents necessary to offer board-level manufacturers a total solution. These software components include the basic input/output system (BIOS) software, drivers for all the popular operating environments and manufacturing test software, as well as the CAD files and supporting documentation (such as bills of materials) necessary for board-level manufacturers to ramp up production of their products as quickly as possible with minimum investment.
Multimedia -- low-cost, mass market multimedia -- therefore, requires a blending of technologies. But in order to provide the best solutions, the integration of functions must be done intelligently, creating synergies among the core technologies, and then developing systems based on these characteristics. This, in turn, requires a company with not only the portfolio of core technologies, but the architectural vision to properly apply them to the PC/Windows environment, the ability to integrate them and the resources to implement products and bring complete solutions to market.
Cirrus Logic is such a company.
# # #