Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing
Figure 6 | EURASIP Journal on Audio, Speech, and Music Processing

Figure 6

From: Deep neural network-based bottleneck feature and denoising autoencoder-based dereverberation for distant-talking speaker identification

Figure 6

Feature space before and after the BF-DNN. We used PCA to reduce the speaker features to two dimensions. The lift graph shows the locations of the different speakers using different colors in a two-dimensional plane. The boundary of each speaker is not clear. The right graph shows the features after the BF-DNN in two dimensions. Here, the boundaries are clearer than those on the left graph.

Back to article page