CycleTransGAN-EVC: A CycleGAN-based Emotional Voice Conversion Model with Transformer
arXiv:2111.15159
Abstract
In this study, we explore the transformer's ability to capture intra-relations among frames by augmenting the receptive field of models. Concretely, we propose a CycleGAN-based model with the transformer and investigate its ability in the emotional voice conversion task. In the training procedure, we adopt curriculum learning to gradually increase the frame length so that the model can see from the short segment till the entire speech. The proposed method was evaluated on the Japanese emotional speech dataset and compared to several baselines (ACVAE, CycleGAN) with objective and subjective evaluations. The results show that our proposed model is able to convert emotion with higher strength and quality.
References in corpus (5)
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Nonparallel Emotional Speech Conversion
- Transforming Spectrum and Prosody for Emotional Voice Conversion with Non-Parallel Training Data
- Emotional Voice Conversion With Cycle-consistent Adversarial Network
- StarGAN-based Emotional Voice Conversion for Japanese Phrases