Comparing objective feature statistics of speech for classifying clinical depression.

Area of Science:

• Speech analysis
• Psycholinguistics
• Clinical psychology

Background:

• Human communication conveys emotional context crucial for understanding mental states.
• Speech analysis commonly uses prosodic and spectral features to classify emotions.
• Glottal waveform features offer potential for emotional state analysis but are less explored due to extraction challenges.

Purpose of the Study:

• To compare major speech analysis categories for identifying and clustering feature statistics.
• To differentiate between a control group and a patient group with clinical depression using speech features.

Main Methods:

• Comparative analysis of speech feature extraction techniques.
• Utilizing prosodic, spectral, and glottal waveform features.
• Clustering feature statistics from control and depression patient groups.

Main Results:

• Glottal waveform features demonstrate significant clustering potential for emotional and stress states.
• Comparison highlights the efficacy of different speech analysis categories in depression detection.
• Identified distinct feature statistics between control and patient groups.

Conclusions:

• Glottal waveform analysis presents a viable, underexplored avenue for depression detection.
• Speech analysis, particularly glottal features, can aid in identifying clinical depression.
• Further research into glottal waveform feature extraction is warranted.