Expression or emotional-motivational connotations with a one-word utterance

Area of Science:

• Speech acoustics
• Phonetics
• Psycholinguistics

Background:

• Understanding the acoustic correlates of emotional prosody is crucial for human-computer interaction and speech synthesis.
• Previous research has explored vocal cues for emotion, but specific phonetic segments like [aa] require further investigation.

Purpose of the Study:

• To investigate the acoustic distinctiveness of emotional expressions in the spoken word [saara].
• To determine listener agreement on emotional categorization of vocalizations.
• To identify specific acoustic features differentiating emotional states.

Main Methods:

• Twelve subjects produced the utterance [saara] with ten different emotional intentions.
• Seventy-three listeners categorized 120 utterances based on emotional content.
• Acoustic analysis examined fundamental frequency (F0), duration, sound pressure, and spectral features of the [aa] segment using Kohonen's self-organizing map.

Main Results:

• High listener agreement (50%-99%) was achieved for most emotional categories.
• Emotions like 'astonished,' 'angry,' 'frightened,' and 'commanding' were generally well-perceived.
• Acoustic features of the [aa] segment, including peak sound pressure and spectral energy distribution, differentiated several emotions, while specific intonations marked 'astonished,' 'scornful,' and 'pleading.'

Conclusions:

• The [aa] vocalic segment contains significant acoustic cues for differentiating emotional expressions.
• Specific acoustic features, such as spectral cues for breathy voice quality, can distinguish similar emotions like 'admiring' from 'content.'
• Vocal emotion recognition is influenced by both prosodic and spectral characteristics of speech.