On incorporating social speaker characteristics in synthetic speech
Sai Sirisha Rallabandi, Sebastian Möller
Abstract
In our previous work, we derived the acoustic features, that contribute to the perception of warmth and competence in synthetic speech. As an extension, in our current work, we investigate the impact of the derived vocal features in the generation of the desired characteristics. The acoustic features, spectral flux, F1 mean and F2 mean and their convex combinations were explored for the generation of higher warmth in female speech. The voiced slope, spectral flux, and their convex combinations were investigated for the generation of higher competence in female speech. We have employed a feature quantization approach in the traditional end-to-end tacotron based speech synthesis model. The listening tests have shown that the convex combination of acoustic features displays higher Mean Opinion Scores of warmth and competence when compared to that of individual features.
Create a lesson
Related papers
A Speech-LLM Framework for Enrollment-Free Wake-Word-Guided Target-Speaker Recognition in Complex Noisy Scenarios for Seamless Human-Machine Interaction
Yuxiang Mei, Dongxing Xu, Yanhua Long
An Interpretable Acoustic Coordinate System for Japanese Consonant Morae in Dysarthric Speech
Yuma Enomoto, Misa Yamaguchi, Hiroki Nishikawa et al.
MOUSSE: Multilingual Overlapped Speech Separation with Predictive Uncertainty Estimation
Su Yeon Shim, Yejin Jeon, David Ifeoluwa Adelani et al.
Auditory-Feedback-Conditioned Flow TTS for Implicit Lombard Adaptation
Yunya Wang, Satoshi Nakamura
Attribution-Guided Shared-Private Decoupling for Noise-Reduced Audio-Visual Representation Learning
Linge Wang, Yingying Chen, Bingke Zhu et al.
ProSing: Zero-Shot Fine-Grained Singing Technique Conversion
Xinhao Guan, Jingyue Huang, Kangrui Chen et al.