📝 Publications

GTSinger: A Global Multi-Technique Singing Corpus with Realistic Music Scores for All Singing Tasks
Yu Zhang, Changhao Pan#, Wenxiang Guo#, et al.

Demo

• GTSinger is a large Global, multi-Technique, free-to-use, high-quality singing corpus with realistic music scores, designed for all singing tasks.
• Our work is promoted by multiple media and forums, such as , , and .

STARS: A Unified Framework for Singing Transcription, Alignment, and Refined Style Annotation
Wenxiang Guo#, Yu Zhang#, Changhao Pan#, et al.

Project |

• STARS is a unified framework for singing transcription, alignment, and refined style annotation based on hierarchical representation learning.

TCSinger 2: Customizable Multilingual Zero-shot Singing Voice Synthesis
Yu Zhang#, Wenxiang Guo#, Changhao Pan#, et al.

Project |

• TCSinger 2 is a multi-task multilingual zero-shot SVS model with style transfer and style control based on various prompts.

A Multimodal Evaluation Framework for Spatial Audio Playback Systems: From Localization to Listener Preference
Changhao Pan#, Wenxiang Guo, Yu Zhang, et al.

Project

• PSA-MOS provides 50 hours of high-quality spatial audio recordings, with detailed localization annotations and fine-grained MOS ratings.
• MESA is a multimodal evaluation framework for spatial audio playback systems which exhibits strong correlation with human perceptual assessments.

ISDrama: Immersive Spatial Drama Generation through Multimodal Prompting
Yu Zhang#, Wenxiang Guo#, Changhao Pan#, et al.

Project

• MRSDrama is the first multimodal recorded spatial drama dataset, containing binaural drama audios, scripts, videos, geometric poses, and textual prompts.
• ISDrama is the first immersive spatial drama generation model through multimodal prompting.

MRSAudio: A Large-Scale Multimodal Recorded Spatial Audio Dataset with Refined Annotations
Wenxiang Guo#, Changhao Pan#, Zhiyuan Zhu#, Xintong Hu#, et al.

Demo

• The largest recorded spatial audio dataset contains four scenarios: daily life, singing, music, and speech, with a total duration of 500 hours.
• Supports multiple spatial audio tasks: audio spatialization, spatial TTA, acoustic event localization and detection(SELD), etc.

Versatile Framework for Song Generation with Prompt-based Control
Yu Zhang#, Wenxiang Guo#, Changhao Pan#, et al.

Project

• VersBand is a multi-task song generation framework for synthesizing high-quality, aligned songs with prompt-based control.