According to many psychological and social studies, vocal messages contain two distinct channels—an explicit, linguistic channel, and an implicit, paralinguistic channel. In particular, the latter contains information about the emotional state of the speaker, providing clues about the implicit meaning of the message. Such information can improve applications requiring human-machine interactions (for example, Automatic Speech Recognition systems or Conversational Agents), as well as support the analysis of human-human interactions (for example, clinic or forensic applications). PrEmA, the tool we present in this work, is able to recognize and classify both emotions and communication style of the speaker, relying on prosodic features. In particular, recognition of communication-styles is, to our knowledge, new, and could be used to infer interesting clues about the state of the interaction. PrEmA uses two LDA-based classifiers, which rely on two sets of prosodic features. Experimenting PrEmA with Italian speakers we obtained Ac = 71% for emotions and Ac = 86% for communication styles.
Extracting Emotions and Communication Styles from Prosody
SBATTELLA, LICIA;TEDESCO, ROBERTO;MATTEUCCI, MATTEO;TRIVILINI, ALESSANDRO
2014-01-01
Abstract
According to many psychological and social studies, vocal messages contain two distinct channels—an explicit, linguistic channel, and an implicit, paralinguistic channel. In particular, the latter contains information about the emotional state of the speaker, providing clues about the implicit meaning of the message. Such information can improve applications requiring human-machine interactions (for example, Automatic Speech Recognition systems or Conversational Agents), as well as support the analysis of human-human interactions (for example, clinic or forensic applications). PrEmA, the tool we present in this work, is able to recognize and classify both emotions and communication style of the speaker, relying on prosodic features. In particular, recognition of communication-styles is, to our knowledge, new, and could be used to infer interesting clues about the state of the interaction. PrEmA uses two LDA-based classifiers, which rely on two sets of prosodic features. Experimenting PrEmA with Italian speakers we obtained Ac = 71% for emotions and Ac = 86% for communication styles.File | Dimensione | Formato | |
---|---|---|---|
extractingemotionslncs.pdf
Accesso riservato
:
Pre-Print (o Pre-Refereeing)
Dimensione
358.17 kB
Formato
Adobe PDF
|
358.17 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.