August 27, 2012
The Computational Intelligence Laboratory (CIL) participated in the speaker clustering challenge organised within the conference “International Conference On machine Learning and Applications (ICMLA) 2010” with the work “Unsupervised Speaker Clustering in a Linear Discriminant Subspace”. The two authors propose a method for separating speakers in the audio signal, using an approximate application of the linear discriminant analysis. The method was evaluated based on data sets of the competition and won the first prize.