Figure 6.
A grid of line plots presents how acoustic features vary with emotion intensity values ranging from 0 to 1 for four emotions: angry, happy, sad, and surprise. Each row corresponds to one emotion, and each column represents a different feature, including duration, pitch mean, pitch standard deviation, energy mean, and energy standard deviation. Within each plot, multiple lines represent utterance level, word level, phoneme level, and combined word phoneme level emotion distributions. For angry and surprise, duration, pitch mean, and energy related features generally increase as intensity increases, while pitch variability shows smaller changes. For happy, pitch mean and pitch variability increase with intensity, while energy features decrease slightly. For sad, pitch mean and pitch variability decrease as intensity increases, while duration shows a moderate increase. Across all emotions, phoneme level features show smaller magnitude changes compared to utterance and word level features.The illustration of prosodic variants with intensity changes. The red background represents the expected negative trend, the blue indicates the expected positive trend, both summarized from the ESD dataset