Have just returned from the annual BCT conference in York. Lots of good talks on offer, as well as lots of interesting trade stands with the latest technology. The Wildlife Acoustics EM3 was popular, giving hand-held outputs of sonograms in the field.

The most interesting talk from my perspective was from Charlotte Walters from the Institute of Zoology who was talking about the results from the iBats project. Charlotte used an ensemble neural network approach (eANN) to try to identify species within the iBats database using parameters extracted from analysis in Sonobat. She managed to get an overall correct classification rate of just over 80% which is comparable to other studies and seems to be about as good as it gets. Myotis as ever were a problem, with Natterer's proving the easiest to identify, with most of the other UK Myotis falling back to about 50% correct classification. The full paper can be found here.