Figure 7.
A set of spectrogram plots presents the effects of emotion intensity control at utterance level and word level for four emotions: angry, happy, sad, and surprise. The left group shows utterance level emotion intensity control, with separate plots for each emotion at intensity 0 and intensity 1, displaying frequency over time with overlaid pitch and energy contours. The right group shows word level emotion intensity control, where specific word regions are modified while surrounding regions remain unchanged. Across all emotions, increasing intensity from 0 to 1 results in observable changes in pitch and energy patterns throughout the utterance for utterance level control, while word level control shows localised changes confined to selected word segments.Spectrograms of synthesized audio samples across different emotion intensities with pitch (blue) and energy (green) contours: the y-axis for energy contours are not relevant. (a) Utterance-level emotion intensity control; (b) word-level emotion intensity control