Figure 5From: Towards an Intelligent Acoustic Front End for Automatic Speech Recognition: Built-in Speaker NormalizationThe variation of the instantaneous self-normalization warp (), averaged self-normalization warp (), and fixed self-normalization warp (obtained from offline BISN w/MS-BTS), the speaker turns are also marked with a dashed line (the averaged self-normalization warp and fixed self-normalization warp are shifted upwards by for proper illustration).Back to article page