Speech

Computers have become quite good at speaking. Computers that speak hide in many devices.

For example, you can buy a talking car that tells you when it needs an oil change, a talking bathroom scale that makes cynical comments about how much your weightís gone up since yesterday, and many other talking devices. You can even buy Coke from a talking vending machine that invites you to deposit your coins and then says "Thank you".

Talking watch

Whenever I want to find out the time, I just press a button on my wrist watch, and its computer voice proudly proclaims the time in perfect English. Whenever I get lonely at night and want somebody to talk to me, I just press the watchís button and thrill to the sound of its soothing voice.

It also acts as the worldís most humane alarm clock. Instead of giving an awful ring, its human voice says, "Attention, please! Itís 7:30AM." Then it plays some jazzed up Bach.

If Iím still sleepy and ignore the alarm, five minutes later it will say, "Attention, please! Itís 7:35AM. Please hurry." It will also subject me to some more Bach. It will keep reminding me every five minutes, until Iím awake enough to turn off the alarm.

You can buy the Vox Watch at Radio Shack for $39.95.

Reading to the blind

The most impressive talking device ever invented is the Kurzweil Reading Machine, which reads books to the blind.

It looks like a photocopying machine. Just lay a book on top of the machine, and the machine reads the book to you, even if the book is laid down crookedly and has dirt on it and has multiple columns and photos and uses weird type.

When it was invented many years ago, it used to cost $50,000. To use such a machine, you had to be rich or live near a library owning the machine. Eventually the price dropped to $20,000. Later, the price dropped even lower.

Now Kurzweil sells a text-to-speech program called Madison for just $50.

It can speak several languages. To use it, you must buy at least a 66-megahertz 486 computer with 16M of RAM and speakers; and if text isnít in the computer already, you must type the text or buy a scanner and a program for optical-character recognition (OCR).

Kurzweil used to be an independent company, but now itís part of Lernout & Hauspie.

Voice input

Though computers are good talkers, theyíre not good listeners. No computerís been invented yet that will replace your secretary and let you dictate a letter to it accurately. But researchers are getting close!

Buy a program called Naturally Speaking (by Dragon Systems, $169) or a similar program called Via Voice (by IBM, $99).

Each program comes in a box that includes a microphone. When you speak into the microphone; the computer listens to what you say and will try to type it for you, so your words will appear on the screen and on disk.

Each program is about 92% accurate: 8% of the words you say will be misinterpreted. You then edit the misinterpreted words. That editing makes the whole process slower than hiring a good secretary who can type fast and accurately.

Because of the errors, neither program should be trusted. For example, if a doctor dictates to the computer that a patient has "irregular heartbeat", the computer might think the doctor said "a regular heartbeat". If a doctor says a patient has "hyperglycemia", the computer might hear it as "hypoglycemia", which is the opposite.

Computer experts love to stump the computer. For example, when Steve Manes said "Hello, Nick and Bob", the computer thought he said "Hello, naked boys" (because "Nick" and "Bob" werenít in the computerís vocabulary yet). When John Woram said "He ate eight oysters", Naturally Speaking thought he said "He 88 oysters"; Via Voice thought he said "Diego State oysters".

Which of those programs is better? Via Voice is better at handling most sentences, but Naturally Speaking is better at handling sentences that include numbers or mouse-movement commands.

Each program requires 32M of RAM. Naturally Speaking requires a 133-megahertz Pentium and 60M of free disk space; Via Voice requires a 150-megahertz Pentium and 125M of free disk space.

Before using either program, you must train the program to understand your accent, by reading some sample sentences.

Music

Computerized music is advancing rapidly. Now you can sit down at a portable piano-style keyboard (light enough to carry in one hand), bang out a tune, feed the tune to a computer, and have the computer edit out your errors, play the tune back using the tone qualities of any instrument you wish (or even a whole orchestra), and print the score on paper.

Such developments are shaking up the entire music industry.

When you watch a TV commercial or movie, the background music that sounds like a beautiful orchestra or band is often produced by just a single person sitting at a computerized music synthesizer. The imitation of orchestral instruments is so exact that even professional musicians canít hear the difference. As a result, whole orchestras of musicians are now unemployed.

Music synthesizers come in two categories. One kindís cheap ($25 to $500) and easy to use but produces sounds that are tinny. The other kind produces beautiful sounds but costs a lot ($500 to $20,000) and is harder to learn to master. Programmers are trying to meld those two categories together. I wish theyíd hurry up!

Ultimate Music Machine

Musicians, programmers, and engineers are working together to create the Ultimate Music Machine, which makes all other musical instruments obsolete. You can buy all its parts at your local computer and music stores, but the software and hardware that connects the parts is awkward. I expect some company will eventually build an assembled version that you just plug into the wall for immediate fun.

Part 1: the tone-quality creator The Ultimate Music Machine can imitate all other musical instruments. To make it imitate an instrument, play a few notes of that instrument into the machineís microphone. The machine makes a digital recording of the instrument, analyzes the recording, and stores the analysis on a 3Ĺ-inch floppy disk.

The machineís analysis is quite sophisticated. For example, it realizes that a violin note has a vibrato (because the violinistís finger wiggles), that each piano note begins with a bang and ends with a hum, and that the pianoís bass notes sound "fatter" than the treble notes (because the bass notes are made from different kinds of strings).

The machine lets you edit the analysis, to create totally new tone qualities, such as "piolin" (which is a compromise between a piano and a violin).

When you buy the machine, it comes with recordings of the most popular instruments, and lets you add your own and edit them. It also lets you use fundamental waveforms (such as sine waves, square waves, and triangle waves), which act as building blocks for inventing sounds that are wilder.

Part 2: the note creator The machine includes a piano-style keyboard (with black and white notes on it). To feed the machine a melody, tap the melody on the keyboard. You can also play chords. The machine notices which notes you strike the hardest, so it records your accents.

The machine includes a pitch-bend dial, which you turn to make the notes slide up the scale, like a slide trombone.

If youíre not good at the keyboard, use the machineís screen instead, which displays a musical staff and lets you move notes onto the staff by using a mouse. You can also use the mouse to edit any errors you made on the keyboard, and to create repetitions and increase the tempo.

If you fear mice and keyboards, just sing into the machineís microphone. The machine notices which notes youíve sung and records them.

If youíre too lazy to create a melody or harmony, the machine creates its own. Its built-in computer analyzes your favorite music, notices its rhythms, note transitions, and harmonic structures, and then composes its own music in the same style.

Part 3: output The machine plays the editing music through stereo speakers. As the music plays, the complete score moves across the screen, in traditional music notation. The machine also prints the score on paper. Yes, the machine prints a complete score showing how you sang into the mike or tickled the keys!

Vendors The Ultimate Music Machine is built from music synthesizers. The most popular synthesizers are made by four Japanese companies: Casio, Roland, Yamaha, and Korg. Their synths cost from $25 to $3000 and contain tiny computers. For extra computing power, attach a Macintosh computer by using a Musical Instrument Digital Interface cable (MIDI cable). To print pretty scores cheaply, add Deluxe Music Construction Set, a Mac program published by Electronics Arts for under $50.

Advanced multimedia

Multimedia is the attempt to make your personal computer overwhelm your senses by feeding you text, music, voice, graphics, animation, and video movies on the screen all simultaneously! To do that well, you need a fast computer (at least an Intel 486 or a Mac 68040) with a CD-ROM drive and circuitry to handle sounds well.

Encarta

Microsoftís most famous example of multimedia is Microsoft Encarta. Itís a CD-ROM disk whose 1997 version includes:

the complete text of the 29-volume Funk & Wagnalls encyclopedia, supplemented by 1000 extra articles (so you get 30,000 articles altogether)

a 65,000-word dictionary

8 hours of sound (organized into 1600 sound clips)

written & spoken samples of 60 languages

7500 photos & illustrations

100 video clips & animations

800 maps

1500 links from the articles to the Internetís World Wide Web

Using Encarta is fun: using your mouse, just click on whatever topic on the screen interests you and ó whammo! ó you see it and hear it. Discount dealers sell it for just $40.

You can also get a souped-up version, called Microsoft Encarta Deluxe Edition, which includes:

two CD-ROM disks (instead of 1)

1800 sound clips (instead of 1600)

14,000 photos & illustrations (instead of 7500)

150 video clips & animations (instead of 100)

4000 Web links (instead of 1500).

Discount dealers sell it for $75.

Cinemania & beyond

Inspired by Encartaís success, Microsoft has gone on to develop other multimedia titles that are more specific. For example, Cinemania is a CD-ROM whose 1997 version includes:

info on 20,000 movies, with reviews by Leonard Maltin, Roger Ebert, Pauline Kael, and Baseline

info on 10,000 directors, writers, actors, & other film folk, including biographies of the 4,500 most important

30 video clips, 150 famous sound bytes ("dialog clips"), and 1000 movie stills

It costs just $20 ($30 minus $10 rebate). I wish the title didnít sound like "sin-o-mania", though most movies are indeed about the manic enjoyment of sin.

Microsoft has also done multimedia titles on topics such as Beethovenís Ninth Symphony (including detailed analysis of the music, the man, and his times), Stravinskyís Rite of Spring, works by Mozart & Schubert, Londonís National Gallery of Art, and baseball lore.

Create your own

Though using multimedia created by companies such as Microsoft can be fun, itís even more fun to create your own!

Most software purporting to help you create multimedia is tedious to use and expensive. But hereís the exception: get Magic Theatre, a CD-ROM disk published by two companies working together (Knowledge Adventure Inc. and Instinct Corporation). Comp USAís been selling it for just $35.

Designed for kids, youíll learn how to use it in just a few minutes. It lets you create animated cartoons with sound, so easily that you can create exciting cartoons after just a few seconds of preparation!

The $35 price even includes a microphone, accompanied by a CD-ROM disk that includes lots of clip art, animated objects, music, and sound effects, which you can combine in just a few seconds to produce an on-screen animated movie that youíll like a lot better than Saturday morning cartoons ó especially since you created it!

The cartoons youíll produce will seem child-like, but thatís their charm!

Try it, youíll like it. If you have kids, the whole family can pitch in to make a family animated movie. Your neighbors will be jealous.