with Prof. J.E. Black
At night, during the spring and fall, millions of small birds migrate over southern Ontario. As they fly they communicate with short call notes or "chips" of about 50 to 100 milliseconds in length. These chips lie somewhere in the frequency range 1KHz to 10KHz. If an automated method of counting the calls and identifying the species could be developed then the counts would provide a valuable estimate of the number of migrants travelling to and from their breeding grounds in northern Ontario. Over the long term it would then be possible to recognize species in decline, possibly as a result of altered environmental factors at the breeding or wintering ground.
In April and May of 1994 VHS recordings were made at St. Davids and Long Point for 8 hours each night. The data was then analysed by listening to the VHS tapes, and the number of chips counted, a long and very tedious procedure. In the fall of 1994 the two stations will again be operated, producing even more data to be analysed. What is desperately needed is computer software for extracting the chips from the tapes and storing them on a disk. The extraction process will give a count of the chips, and the stored data can then be analysed to determine the species of bird that made the call.
The essence of the problem here is that of extracting the bird chips from the noise. On some nights the noise level is low, but on others wind, rain, automobiles, waves (at Long Point), birds calling on the ground, dogs, raccoons, squirrels and many other things may contribute to the noise background. Somehow the human ear has theability to detect even the weaker chips in all this noise. The objective of this project would like to build computer software that can duplicate this ability.