Singing voice synthesis based on frame-level sequence-to-sequence models considering vocal timing deviation

Re-assign community