Improving speaker verification robustness with synthetic emotional utterances

Open in new window