Thank you

[ChuxiJ]

Thank you very much for your contributions to the evaluation of the song! I've been searching for a music generation evaluation solution similar to TTS, but haven't found a best practice yet. Your evaluation method is incredibly detailed. While it's still in the early stages and has room for improvement—such as alignment between style and instructions, alignment between lyrics and content, and addressing style/language biases—it's already a fantastic starting point. Thank you sincerely!

[yaoxunji]

Thank you very much for your contributions to the evaluation of the song! I've been searching for a music generation evaluation solution similar to TTS, but haven't found a best practice yet. Your evaluation method is incredibly detailed. While it's still in the early stages and has room for improvement—such as alignment between style and instructions, alignment between lyrics and content, and addressing style/language biases—it's already a fantastic starting point. Thank you sincerely!

Thank you so much for your kind words and thoughtful feedback! I'm really glad to hear that you found the dataset and evaluation method helpful. You're absolutely right—there's still much room for improvement.
I hope this work can serve as a foundation for further research in song evaluation. Please feel free to contribute or share ideas—I'd love to collaborate or discuss further!