Towards Improving the Expressiveness of Singing Voice Synthesis with BERT Derived Semantic Information

Open in new window