Fig 15. Classification results (mean value of the 10-fold for error rate, sensitivity and specificity) for all channels of speaker 1.

These results are based on four additional sessions from this speaker recorded a posteriori. Error bars show a 95% confidence interval.