Digital Domain -- (By Ken Pohlmann; Feb. 1987)

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting


Departments | Features | ADs | Equipment | Music/Recordings | History



DISC OF MANY (INTER) FACES

As we observed last month, the CD-I format opens diverse new applications for the Compact Disc storage medium. Its ability to simultaneously store audio, video, text, and data could create a wholly new publishing medium, read through the tiny laser beam in your new CD-I player. Let's continue our introduction to this new format (with thanks to Philips International's helpful text, "A General Introduction to CD-Interactive," which provided much of the information in this and last month's columns).

To make room for video and text, CD-I must compress the audio information which otherwise would occupy the entire disc space, as it does in the CD-Audio format. Adaptive delta pulse-code modulation (ADPCM) accomplishes the feat. ADPCM is an efficient variation on pulse-code modulation (PCM) and is extremely useful because of its ability to store digital audio data with fewer bits. On the other hand, ADPCM requires more processing than regular PCM for both encoding and decoding.

Pulse-code modulation encodes the absolute value of each signal sample as a digital (usually 16-bit) "word." Delta modulation encodes relative values, comparing each sample to the previous one and using a one-bit signal to indicate whether the next level lies up or down the digital audio staircase.

This saves a lot of data space, but it encodes signal-level changes only in fixed amplitude increments for each sample. With adaptive delta modulation, the amplitude of the step size may be varied to more closely approximate the waveform. In ADPCM, a PCM word is substituted for the one-bit correction signal. Now there are several quantization levels available, each with its own step-size scale factor. The result? Even better audio performance, with a relatively small number of bits.

The ADPCM used in the CD-I format uses four- or eight-bit PCM words, de pending on the level of sound quality required. With a four-bit system, correction information is available at one of 16 levels. An eight-bit word provides 256 levels, for even better performance, at the price of more bits. Each of the quantization levels is assigned a step-size scale factor; because considerable step-size information is avail able, step sizes may be adapted with great accuracy. The scale factors themselves are based on the statistics of the signal itself. For example, scale factors for an ADPCM circuit designed to process speech would be selected differently from those for a system de signed to process music. The CD-I for mat calls for three levels of ADPCM fidelity: "Hi-Fi" (LP quality), "Mid-Fi" (FM quality), and "Speech" (AM quality). In addition, regular 16-bit PCM audio can be encoded. By using efficient ADPCM for the lower fidelity modes, more space is made available on the disc for accompanying video or computer data, or for more channels or more recording time.

PCM recordings, like CD-Audio, are stereo, and fill the disc. The three ADPCM modes, however, can be stereo or mono. The number of possible channels increases as the fidelity level decreases, as shown in Table I. Of course, there are always twice as many monaural channels as stereo channels. For example, in Mid-Fi mode, there could be four stereo channels or eight monaural channels. In Speech mode, a disc could have up to 16 72-minute monaural channels.

There is a pause of 1 to 3 seconds when switching from the end of one channel to the beginning of the next.

The audio quality level chosen deter mines how much storage capacity is left for video information. As Fig. 1 shows, a disc can be all audio or all video, split 50-50 (using Hi-Fi stereo sound) or divided to give more space to video by using the Mid-Fi or Speech audio quality levels. Using Speech mode, up to 94% of the disc space is available to hold video or other non-audio information.

Table I--Levels of CD-I audio fidelity and resultant channel availability.



Fig. 1--Comparative video and audio information densities for CD-I. By using lower fidelity, less information-intensive audio modes, more room is made available on the disc for video (or other) information.

A CD-I disc can store video material with varying quality levels for resolution and pixel coding. Two standards of video resolution are supported: Normal resolution of 384 x 280 pixels (picture elements) and high resolution of 768 x 560 pixels. Normal resolution corresponds to best achievable resolution with normal television receivers, while high resolution is the best that's likely to be achieved with future enhanced or digital receivers. Pictures are generally non-interlaced, although interlacing can be used.

The CD-I specification for pixel coding provides for three picture qualities:

Studio (natural) quality and two graphics levels. Natural pictures normally occupy about 325 kilobytes per picture without interlacing (650 kilobytes with interlacing). However, the CD-I system uses a data-compression technique to reduce storage requirements to 108 kilobytes. At a data rate of 174.6 kilo bytes per second in CD-I's Form 2 configuration, one full-frame natural picture is transferred in just over 0.6 second. That's no competition for videodisc or videotape, which transfers one frame every 0.033 second.

The first of the two graphics modes is designed for applications that involve image manipulation by the end user. This mode is based on absolute RGB (red-green-blue) coding. It supports either eight-bit color coding, yielding 256 different colors, or 15-bit coding for 32,768 colors. A 15-bit RGB graphic would occupy about 215 kilo bytes per picture. No compression is used in this mode.

The second graphics mode, de signed for animation, is based on color look-up table (CLUT) graphics. This mode permits full-screen animation with four-, seven-, or eight-bit coding for 16, 128, or 256 colors, respectively.

The eight-bit CLUT mode requires 108 kilobytes per frame, but compression can reduce this to typically less than 10 kilobytes per frame. With compression and the interleaving of sound and picture, this graphics mode can pro vide animation at a rate of 17 frames per second. That's about the same frame rate as silent films (16 or 18 frames/S) but slower than sound movies (24 frames/S) or television (25 frames/S in Europe, 30 frames/S in the U.S., Canada, and Japan).

Text can be stored with bit-map or character encoding. The bit-map pro cess, which draws pictures of each character, can handle any character (including math symbols, foreign alphabets, astrological symbols, or miniature smile buttons), but requires five bytes per character for storage. With the bit-map process, if the characters are drawn in an 8 x 10 matrix of dots in 16 colors, a disc can hold a maxi mum of 120 million characters. These characters cannot be manipulated under program control. However, it is possible to place them electronically onto transparent or translucent over laying planes. The system specifically allows for superimposing bit-mapped text onto external video, or pointing at portions of the text with an externally controlled cursor.

Character-encoded text can handle only characters for which codes have been standardized. This mode re quires only one byte per character for storage, thereby limiting the number of different standard characters to a max mum of 256 but allowing storage of up to 600 million characters on a disc.

Text can also be encoded with two bytes per character, with the second byte specifying such factors as color, font type, and size. This would allow a total of 300 million characters per disc.

In both character-encoded modes, it is possible to manipulate text via soft ware-copying text from the disc into a document being written on the user's computer, for example.

In the normal-resolution mode, text is limited to 40 characters on 20 lines to allow display on ordinary TV sets. The high-resolution mode allows 80 characters on 40 lines for display on computer monitors (which currently handle only 25 lines of 80 characters) and future digital TVs. A wide range of visual effects is defined, including cuts, scrolls, overlays, dissolves, and fades.

CD-I is not a peripheral but a self contained system. To ensure universal disc and drive compatibility, dedicated hardware and interfaces are specified.

A CD-I player will contain a disc drive, decoder chips (for text, graphics, video, and audio), and microprocessor controllers; it could be interfaced to your television and stereo. Although a CD-I player could also be interfaced to a personal computer, it would not be cost-effective if used for that alone, and it would also miss the medium's intent. A CD-ROM drive and interface alone would make a better computer peripheral.

The CD-I system uses the Motorola 68000 microprocessor family. The CD-I real-time operating system (CD RTOS) is based on the OS-9 real-time operating system. This ensures that all CD-I players will be able to search, retrieve, process, and output any information stored on any CD-I disc.

Because the CD-I format recognizes 16-bit PCM data, a CD-I player will be able to play regular CD-Audio discs. A CD-Audio player will not be able to play CD-I discs, however. As specified in a proposal from the High Sierra Group (an industry committee), CD-ROM Mode 1 discs can operate on CD-I systems. On the other hand, a microcomputer with a CD-ROM drive is not always capable of processing the information on a CD-I. Hopefully, universal players will appear, able to play any kind of CD: CD-Audio, CD-I, and all CD-ROMs.

The standards for CD-I disc layout and file structure specify a number of criteria. A CD-I disc, even if it's other wise devoted to pure audio, must still begin with one track of CD-I information. This track, even if it contains nothing else, must contain at least the disc label information, in block zero. The second and subsequent tracks may then be used for CD-Audio or other data. In this way, if a CD-I disc with 16 bit PCM audio information is played on a CD-Audio player, the CD-I track can be skipped. Conversely, since regular CD-Audio discs may be played on any CD-I player, the player must be able to recognize these discs, which have no initial CD-I track, for what they are.

Ir, addition to identifying the disc (alone or as part of a collection), the CD-I label contains information about the disc's type and format. The label also gives information about the locations of file directory information and of the bootstrap program required to get the system going.

Obviously, the scope of the CD-I specification is considerable. The system's designers anticipated that a number of diverse players with different performance features and levels would be developed. Thus, to ensure basic compatibility, the CD-I specification provides a minimum set of requirements for CD-I systems.

A single CD-I disc might contain 19 hours of audio, 41,000 color pictures, 300,000 typed pages, or any combination of the three. Thus, CD-I applications promise to be diverse. A CD-I dictionary, for instance, might contain a word and its definition, as well as spoken pronunciation, pictures, and additional cataloging of synonyms, antonyms, word relationships, origins, or translations into foreign languages.

Another application area is the "teach-yourself" or "how-to-do-it" field; the CD-I's ability to convey text, pictures, and diagrams combined with sound makes it ideally suited. For example, the sound of a motorcycle engine in various stages of tuning could be re produced on the disc, or a text on ornithology could reproduce bird calls. Tourists could obtain a multi-media preview of favorite vacation spots. Works of fiction could be provided with a labyrinth of plot deviations which would change the story each time the text is read or which could be steered at the discretion of the reader himself.

The CD-I format presents consider able opportunities for the hardware, software, and publishing industries to provide consumers with new forms of interactive entertainment and education. However, there are not many existing, readily transferable programs that take advantage of the video, graphics, audio, and text capabilities supported by this new standard. There is a considerable amount of creative work ahead for home use.

When will it all happen? CD-I proponents (Philips and Sony) are projecting U.S. introduction late this year. Of course, it will take some time for extensive CD-I software to become available.

And what about the good old CD Audio disc? CD-I players will play all existing music CDs with the same fidelity as CD-Audio players and, of course, do much more. In that respect, CD-I will be the ultimate CD system.

Still, I feel a little sad. Part of the reason that the Compact Disc is so likable is because it is so simple. Now, with CD-ROM and CD-I, everything suddenly gets a good deal more complicated. Well, that's progress.

(adapted from Audio magazine, Feb. 1987)

Also see: Dr. Thomas Stockham on the Future of Digital Recording (Feb. 1980)

= = = =

Prev. | Next

Top of Page    Home

Updated: Saturday, 2019-05-04 12:57 PST