Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing
Figure 2 | EURASIP Journal on Audio, Speech, and Music Processing

Figure 2

From: Perceptual audio features for emotion detection

Figure 2

Frame-based perceptual feature extraction. In order to efficiently model the characteristic of an emotional mode, the perceptual features computed in each frame are averaged over Y overlapping frames. Hence, a single M-dimensional average feature vector is extracted from each emotional and reference audio pair and is used as a training or test vector.

Back to article page