A Cochlear implant user participating in an instrument recognition exercise. Her processor is a ‘behind the ear’ model.

A new patent awarded to researchers from UNSW and Cochlear Ltd is designed to improve the perception of music—and also of tonal languages—by cochlear implant users. The new processing strategy was developed and implemented using results from the doctoral thesis submitted in December 2001 by Robert Fearn, who is one of the authors of the patent.

A cochlear implant uses a microphone, a signal processor, a radio link from external to internal electronics and an implanted array of electrodes to deliver series of electrical pulses to different regions of the auditory nerve. It works well for speech, with many users being able to conduct normal telephone conversations. However, it works poorly for music, with the result that many users turn them off when music is played.

For his doctoral research Fearn studied the perception of music and related stimuli by cochlear implant recipients. Among other things, his thesis provides a mapping between parameters of the electrical signals injected and perceptual parameters reported by the users in psychophysical experiments. Perceived quantities including pitch, resolution, jitter and quality of sound were determined and related to the signal parameters, including the electrode position and stimulation rate. At low frequencies, the pitch depends approximately logarithmically on stimulation rate, but saturates at high frequencies. Over all frequencies, it depends roughly linearly on the distance of electrode insertion in the cochlear.

The patent, of which the authors are Joe Wolfe and Robert Fearn at UNSW and Paul Carter, Simon Parker and Niki Frampton, researchers at Cochlear, is for a new coding strategy designed to deliver efficiently the selected combinations of pulse trains with the ‘right’ stimulation rate to the ‘right’ place, so as to improve the perception of pitch and harmony, and the naturalness of the sound. Fearn’s laboratory tests using synthetic signals have shown how pitch resolution and sound quality are improved by the new strategy. Model experiments also gave reason to expect improved perception of tonal languages such as Chinese.

Fearn now works for Cochlear, where he has installed two different versions of the strategy into a series of CI processors that are undergoing preliminary trials with volunteer users. Clinical trials begin in 2002, and the early feedback is encouraging.

Robert Fearn and Joe Wolfe