The Future of Voiceovers: Hold Your Tongue...Possibly Forever

"Do we need to cast a voice-over talent for this project?"

That's a valid question any producer might ask when creating an advertisement, corporate audio-video presentation, video game, etc. Of course, the answer depends on what elements the producer and client feel will best communicate with the audience.

For a radio ad, a fully sung jingle with no voice-over could work best. A TV spot or corporate narration might be most effective using scrolling graphic and text, again without an announcer. But very soon producers could be pondering whether their productions need a voice over talent for a more disturbing reason. Vocal utterances produced by air passing through folds of tissue and formed by lips, teeth, and tongue may, simply put, become obsolete. Yes, the "virtual voice talent" may very well become a reality.

Welcome to the Machine

In the May 2004 issue of Mix Magazine, in two separate articles, Stephen St. Croix and Paul D. Lehrman relate their experiences with a new piece of software ominously named "Vocaloid." This little computer-coded wonder is a speech synthesizer that's being used to synthesize background vocals on actual recordings that are being sold to the public--background vocals so good, you'd be hard-pressed to recognize they're fake singers. Now, considering the dubious singing talents of many of our current pop stars, maybe a Vocaloid virtual diva named Britney isn't too far-fetched. Audio manipulation, including pitch correction, equalization, compression, reverb, have been used for decades to save the bacon of many a pop star's performance in-studio or on stage. Technically, it's just a short step from this point to a "singer in a box."

In fact, in the letters section of the July 2004 issue of Mix Magazine a person identified only as "BC," referring to the St. Croix and Lehrman articles, boasted that he's created a "band" called The Bots, "...created wholly from speech synthesizers and 3-D graphics." BC further states, "I use Vocaloid among a variety of other speech synths to make it more into an ensemble. The Bots have released two CDs, a 'record deal' with Magnatune, and a second video in the works. It's been a long and painful ordeal, but I've finally gotten them to the point where they seem as real as any other band out there--except no live concerts."

I've Gotta Sing

And that's the crux of the matter. The appeal of virtual entertainers probably will be quite limited--at least for the foreseeable future--because they can't tour, do drugs, get into fights, sue their record labels, promote world peace, raise money for charity, or do anything live flesh and blood performers can do. We, the audience, love the performer as much as the performer's music. And, in this case, that's a good thing. Tony Bennett, the White Stripes, Diana Krall, Toby Keith, Frederica von Stade, and all of the American Idol wannabes are quite safe from Vocaloid elimination.

Speak Now of Forever Hold Your Peace

But voice over talents may not be so lucky. Voice talents are not seen. They don't have adoring fans, except their moms and, maybe, a few other voice-over talents. They perform in short increments: 30 seconds, 60 seconds, a 30 minute narration on how to make a million in real estate. If speech synthesis has reached a point of sophistication sufficient enough to create virtual singers, what's to prevent a software genius from developing a program to replace voice talent? Write the program.

Sample 300 to 500 voices, male and female, each with unique characteristics, incorporate them into the software and, voila, Instant Announcer in a Box. Just load your script text into the program, which converts the text to perfectly uttered speech. No retakes. With a few clicks of the mouse to tweak inflection, emphasis, pacing, dynamics, etc. to polish the natural feel of the voice-over and you're done.

Far-fetched? If entertainment's got a virtual band call The Bots, why can't advertising and marketing have its own virtual Don Pardo?

Well, it seems maybe they can...