Singing Voice Synthesis
Introduction
AI music uses artificial intelligence models to analyze, generate, and transform musical elements such as melody, harmony, rhythm, and timbre. By learning from large collections of audio and symbolic music data, AI systems can compose original pieces, assist human creativity, and perform tasks like music style transfer, accompaniment, and audio restoration. Today, AI music is widely applied in content creation, games, film scoring, education, and personalized music experiences.
Founded in 2025, our research group focuses on building high-quality music datasets for AI and tackling key challenges in music and audio intelligence. Our work addresses problems such as deepfake music detection, singing voice conversion, and singing voice synthesis, aiming to improve both the reliability and creative capabilities of AI music systems. Through data-centric research and advanced modeling, we support robust, ethical, and expressive applications of AI in music.
Contact: Dr. Nguyen Thi Thu Trang | ✉️ trangntt@soict.hust.edu.vn
Research Direction
- Singing Voice Dataset: While there are extensive datasets available for widely spoken languages such as English and Chinese, resources for less common languages like Vietnamese are still scarce. One of the main reason is the difficulty in collecting the clean and high-quality singing vocal and music annotation. Our main research direction for this area is to build a fully-automatic pipeline to collect, clean singing vocal from public sources and musically annotate using deep learning methods.
- Singing Voice Dataset: Our research group focuses on the complex challenge of Singing Voice Conversion (SVC) within mixed acoustic environments. While traditional SVC models often require clean a cappella inputs, our work targets the disentanglement and synthesis of vocal characteristics directly alongside instrumental accompaniment. By leveraging advanced deep learning architectures, we aim to achieve high-fidelity vocal transformation that preserves the original musical context, rhythm, and emotion.
Team Members
Vu Duc Minh
Team Leader
Nguyen Huu Cong
Researcher
Pham Duc Phuoc
Researcher
To Duy An
Researcher