0:00:15an initial draft and that would like to well
0:00:18a given introduction to the to the work done by olivier models on all of
0:00:22the s two cannot be here
0:00:25and try to be quick
0:00:28so of the work is about some analysis homework a sorry forgot the title skull
0:00:33analysis an optimization of bottleneck features for speaker recognition
0:00:37and it's basically are about features that are used as an input to the to
0:00:42the widely used bottleneck features
0:00:47so first we started by using the of the sr features and try to optimize
0:00:52the network to give the best phoneme recognition for phoneme error rate basically
0:00:58and what we see that it is that she try to put in the features
0:01:02that we know there are best for these speaker recognition which is the mfccs plus
0:01:07some extra normalisation plus some extra constraint on the on the neural network training et
0:01:14cetera so the conclusion is that the asr features
0:01:18are not the best for of speaker recognition using the bottleneck ubm i-vector system
0:01:24and better phone accuracy does not necessarily mean a better speaker error rates so thank
0:01:31you once the poster