2012年1月15日星期日

A pitch-perfect pitch at CES

Concert pianist Bob Taub watched his teenage daughter strive to learn the violin, and wondered: Could he design a digital tool to help her see and hear the exact notes where she was going astray?

Last week, six years into an add-on career as an inventor, Taub was here at the massive International Consumer Electronics Show, moving his idea a bit closer to reality.

His invention, MuseAmi, is already at the heart of an iPhone app that may prove to be the ultimate Karaoke tool. You sing into your phone, and ImproVox corrects your pitch, adds harmonies, and even mixes in effects such as echo and reverb.

But that is just the start for a software engine so powerful that it can read a musical score and translate it into sound, or listen to sound and transform it into a musical score.

As Taub, who lives in Princeton, sees it, MuseAmi could transform how people learn, sing, record, and perform all kinds of music. And it could eventually be at the heart of devices that will hear words spoken in one language and play them back - in real time - in another.

Big ideas are the lifeblood of the Consumer Electronics Show - an event that's known to geeks and engineers everywhere simply as CES and that for a few days each year turns Sin City into Technopolis.

CES is a glitzy stage and showcase for many of the biggest names in technology. But it is also a magnet for budding entrepreneurs, inventors, and start-ups that want to change the world, or at least a small slice of it, and perhaps make a fortune along the way.

Some, such as FreeOneHand, are here to show off inventions that, strictly speaking, are not electronic at all but capitalize on devices such as the iPhone and iPad. A spin-off of a Northeast Philadelphia injection-molding company, Engineered Plastics, FreeOneHand developed an ergonomically improved iPad holder. With the $40 device, the company says, an iPad user can read for hours without the usual fatigue.

Others, such as Chalfont-based Telikin, are trying to carve a niche directly into the businesses of industry giants. Telikin makes two simple-to-use touch-screen computers geared to customers, such as senior citizens, who might be scared away by the complexity of a Windows computer or the price tag of an Apple.

Based on the open-source Linux operating system, the $699, 18-inch Telikin Touch and the $999, 20-inch Telikin Elite offer virtually everything a home user might want, including access to e-mail, the Web, photos, video, and documents. They also come with features especially appealing to newbies - or to children or grandchildren who might want to buy one for a family member.

One feature, Tech Buddy, allows the computer's owner to grant remote-control access to a relative or friend - say, in case Mom needs help or a refresher lesson. Another is a special link that shows Facebook photos without a visit to the social network's website.

"We have a lot of baby boomer customers who have parents who want to see pictures but are leery of Facebook," said sales manager Cheryl Lewis.

A large part of what matters at CES happens behind the scenes, however - the place where Bob Taub and a colleague met with representatives of some of the biggest names in the electronics and music businesses, although he is barred from naming names by nondisclosure agreements.

One clue came from where Taub camped out during his days at the show: at the booth of Qualcomm, a leading chip-maker that makes the brains for iPhones, Androids, and other cutting-edge devices.

The booth displayed a demonstration product that MuseAmi created with Qualcomm, which has technology that allows two people using separate microphones to participate together in an activity. The demo was a game in which players compete to identify a song when a brief snippet is played - similar to a one-player game, Drop the Needle, that MuseAmi has offered since the summer as an iPhone app.

Taub said the apps were just a taste of what MuseAmi can do, thanks to elegant "machine learning" algorithms developed by an impressive team that includes Yann LeCun, a professor of computer science and neural science at New York University and a leading developer of a kind of artificial intelligence known as "convolutional neural networks."

LeCun's networks play a key role in software that can distinguish between a genuine signature and a forgery with nearly 100 percent accuracy. Bob Stockman, chief executive officer of a biotechnology company, Reva Medical Inc., and MuseAmi's lead angel investor, agrees with Taub that they can play a similar role in machine learning of music and voice.

"The largest music studios, and people developing chipsets for computers and handsets and businesses, see the power of this technology," Stockman said. "The demonstrations blow everybody away, and they all want to figure out a way to incorporate it."

Taub's musical prowess certainly helps. A graduate of Princeton University and the Juilliard School, he has performed since the 1980s with the Philadelphia Orchestra and other leading orchestras around the world.

So, undoubtedly, does a lifelong fondness for tinkering with things, such as the vintage motorbikes he has rebuilt.

Taub said MuseAmi analyzes six parameters of music: frequency; rhythm; note clarity; volume; and vibrato. Its intelligence is so sophisticated that it should be able to identify a song even when a bad singer is attempting it, he said.

But MuseAmi's most impressive feature may be the simplicity of its software. Taub said ImproVox, which contains many of MuseAmi's key features, uses only about 20 percent of an iPhone's processing power.

Even with that, it is powerful enough to fix your pitch, create harmonies - even build four separate lines, so you can sing with yourself like a barbershop quartet.

没有评论:

发表评论