BOFFINS AT ROCHESTER UNIVERSITY say they have managed to digitally recreate a 20-second clarinet solo and cram it into less than a single kilobyte - 1,000 times smaller than a regular MP3 file.
The achievement, unveiled yesterday at the International Conference on Acoustics Speech and Signal Processing in Las Vegas, was done by first reproducing real world physics of a clarinet by a computer and then modeling the actual physics of a clarinet player, digitally replicating the original performance using everything a computer has been programmed to know about clarinets and clarinet playing. The results, although not yet flawless, are pretty close to the real deal.
Mark Bocko, professor of electrical and computer engineering and co-creator of the technology reckoned that the new technology was "essentially a human-scale system of reproducing music".
Xiaoxiao Dong and Mark Sterling, both doctoral students who worked on the project, said they had examined every single facet of what goes into making up a clarinet’s sound, including backpressure in the mouthpiece, various fingerings, and the way sound emits from the instrument. With that information, they then proceeded to build up a computerised clarinet and virtual player for the digital instrument.
To do this, the students had to observe how real life clarinet players related to their instrument, from simply working out the possible fingering combos, to the force of breath they needed to use and the pressure exerted by the musician’s lips. All this was taken into account in a model which was then able to know how each aspect alone would affect the virtual clarinet.
Once all of that was out of the way, the researchers let the computer " listen" to a real life clarinet recital, and digitally deduce and log the different actions needed to form a particular sound. The original sound could then be recreated as the record of the player's actions were fed back into the computer model.
"Maybe the future of music recording lies in reproducing performers and not recording them," said Bocko, who also reckoned that although not yet tested, the system could probably be extended to vocals too.
Bocko’s boffins are also working on ways to combine more than just one instrument at a time. What they have managed to do, with help from post-doctoral researcher Gordana Velikic and Dave Headlam, professor of music theory at the University of Rochester's Eastman School of Music, is create a way of extricating multiple instruments from a mix enabling highly compact recordings.
The researchers are also working on bettering their acoustic measurements in order to produce more and more precise synthesis algorithms allowing for maximum data compression of music. µ
