1.5-Gigabyte Piano Kills the Chipmunk Effect

A new technology from Rockwell International puts seven channels of audio and seven velocity layers into each of the 88 keyboard notes. Digitally sampled instruments never sounded so grand.

A new technology announced by Rockwell International's Brooktree Division, makers of high-performance integrated circuits for multimedia, could vastly improve the sound of digital musical instruments.

At the National Association of Music Merchants on Thursday, the company unveiled its new Endless Wave technology, which would give PC-based digital instruments real-time access to unlimited sample libraries on mass storage devices such as hard drives.

Until now, most sample-based musical instruments use one sample of a sound, which is "pitch shifted" to allow that sound to be played up and down the length of a keyboard. This does not accurately reflect the range of an acoustic instrument's qualities when playing different notes.

To demonstrate the advantages of its new technology, Brooktree has created what it's calling the "largest sample instrument ever" - a grand piano based on seven channels of audio and seven velocity layers for each of the 88 keyboard notes. The instrument very closely replicates the characteristic of all notes, no matter how hard they're struck.

"That's a smart way of doing it, isn't it?" asks Kenneth Newby, composer, performer with Trance Mission, and instructor at Simon Fraser University's School for the Contemporary Arts.

"A piano is like 88 separate little instruments," said Newby. "Until now, you've had to deal with the 'chipmunk effect' which is when a sample just loses it's accuracy at different ends of the pitch scale. Although I don't use piano sounds much, I could see using something like that - that amount of memory at your disposal would be like having a whole orchestra."

Newby sees the technology's limitations surfacing with more complex instruments like the saxophone, which have much more variable human interaction. "For that, you need more complex physical modeling," he said.