Emotional Voice Reconstruction Based on Intonation Alteration

Abstract

Emotion recognition and voice emotional alteration are major topics in the field of artificial intelligence. Their possible applications include driver’s physical state detection, human-machine interaction, monitoring of students, voice training and robotics. There are many models to express a wide variety of emotions felt or expressed by humans. While there is much work regarding emotion detection, our work will focus on creating emotional voice by altering intonation and rhythm by methods such as overlap-add. We succeeded in detecting an emotional intonation pattern and reconstructing it on a same phrase said by a different voice.