Synthesized Singers

ctwxman writes "Over the past few decades, advances in computer hardware and software have eliminated many jobs... some technical, some menial, but none artistic. As an on-camera performer in television, I've always was believed that I was 'bulletproof' as far as replacement through technology was concerned. Not so fast. Recently, The Sinclair television stations began using 'central casting' to bring news and weather anchors from a central location (near Baltimore) to the local outlets. Still, real people are needed, just not as many. But now, even real performers may be replaced. The New York Times (inhalation of airplane glue required) reports on a new technology which allows synthesized singers to sing. Imagine having a singer with a world-class voice at your disposal, any hour of any day. She's just standing at the ready, game to perform whatever silly song you might make up for her: a ballad about her love for you, a tribute to your best friend's golf game, a stirring rendition of the evening's dinner menu. Scary."

Comments

This isn't really NEW

[Arker]

It sounds like they've gone to much greater lengths on this project than any I'm aware of in the past, but the basic thing here has been out for a long time. Most any keyboard you can buy has human voices. A single sample can be spread out over your keyboard and sing any pitch you want, even glides and stuff, pretty easily. But it's generally fairly rudimentary - 'ahh' and 'ohh' or similar, you can actually do some nice sounding background vocals but not sing verses.

From the description in the article, this 'new' thing is really just an inevitable extension of that - they spend about 5 days with a singer, recording her singing many different phonemes and different effects, so that you can then piece together the words to your own song and put it to your own melody in her voice. And, for the moment, they're still aiming at producing background vocals, just more complex ones with the ability to do actual lyrics instead of a oohs and aaahs. Could be kind of cool, but it definately doesn't sound like a 'quantum leap' - just an extension of long-existing technology. I've been expecting to see someone do this for well over 10 years now, ever since I first got to play around with a digital synthesizer.

Re:This isn't really NEW

[iabervon]

It might actually be easier to do singing than normal speech, because singing replaces intonation, tempo, and some of stress, all of which otherwise have to be determined from a syntactic and semantic analysis of the text in order to really sound right. There have been people who have learned to sing songs in languages they didn't know at all, while I have yet to hear of someone giving a lecture (as convincingly) in a language they didn't know.

Re:Scary?

[cgranade]

This isn't about tech. It's about the need for human creativity and artistry being diminished. I, as a geek, like tech to the extent that it reduces the tedium and frees us to be creative. This is realizing that the very thing we love can be used to work against us. And that is the realization that is truly and deeply scary.

Re:Scary, or progress?

[Anonymous Coward]

What is wonderful about this is that (initially at least) it will devalue the type of generic boy/grrl band trash music which so saturates the current pop market. That's got to be scary for organisations like the RIAA, who actively market the interchangeable swill.

When pitch perfection and standardised voices are available from a $300.00 software package, music made by people with interesting voices and offbeat musical philosophies will be that much more valuable.

After all, it seems unlikely that there'll be a software Tom Waits or a digital Johnny Rotten in our immediate future. Punk revival anyone?

There are more artists than performance artists

[LuxFX]

Over the past few decades, advances in computer hardware and software have eliminated many jobs... some technical, some menial, but none artistic

Ever hear of a cel animator?

Re:Not really buying it.

[Reteo Varala]

I can see where you're going with that argument, and to be quite honest, I don't put much faith in AI, either. The best example of what I think about it is based in an old Infocom game, "A Mind Forever Voyaging."

Artificial intelligence isn't truly artificial sentience until it has the capability of experiencing it's own existance. Living organisms that posess such self-awareness have thousands of input devices, known as nerve receptors, which alert them to the presence of anything to their immediate position. By this, one must learn to recognize the receptors' data. After a long time of learning the abilities of those receptors, and their cousins, the motor nerves (which activate muscle groups for the purpose of movement), self-awareness becomes available, because everywhere on the human body has such receptors, and what doesn't isn't really the human body.

With this knowledge, the person then begins to learn what is a pleasant experience to those receptors, and what is pain. With pleasure/pain, over time, the person begins to develop affections and apprehensions, which give way to full emotional response. Some additional functions in the body help this along, such as endorphins which improve the pleasure state in the brain, and thus, the body... further enhancing the personal experiences.

Now, a computer would have to have MASSIVE amounts of electric and processing power to activate and stimulate such receptors, should miniturization ever allow such devices to be manufactured cheaply and at such quantity to compare with the human's nervous system. And without that system, a computer cannot develop the deep, intricate levels of affection/apprehensions that would allow for emotional responses.

Add to this the fact that a computer would have to be able to process all of this in realtime, over approximately 12-18 years to truly mature into a true artificial sentience.

Now, what does this have to do with music?

Music is all about experience. People write what they know, and they sing how they feel. Experience is a byproduct of sentience, which most definitely means that computerized music, which can please and FOOL audiences, is yet a long time in coming.

Walk before you run, talk before you sing

[shirai]

Considering that normal speech synthesis has not been done well, singing seems to be hard. Already people can take a bad singer and turn them into a good singer but complete synthesis seems unlikely.

Furthermore, this tech is likely not going to be what you think. What makes a singer good is their INTERPRETATION of the notes. Even with proper synthesis, at its best, it will be like computer animation. It could be very good and maybe even perfect but it would be TIME CONSUMING. Watch the making of Making Nemo on the DVD to get an idea of how hard it is to understand emoting.

You would really need to spend a large amount of time figuring out how to make the voice sound EMOTIONAL.