ComfyUI Node: YuE2 生成乐谱计划

YuE2 生成乐谱计划:

The YuE2PlanSong node is designed to facilitate the creation of a symbolic plan for song generation using the YuE2 framework. This node plays a crucial role in the music generation process by taking user-defined inputs such as musical style and lyrics, and transforming them into a structured plan that can be further processed to generate music. The node leverages advanced algorithms to interpret the style and lyrical content, allowing for the creation of a detailed musical blueprint. This blueprint serves as a foundation for subsequent stages in the music generation pipeline, ensuring that the final output aligns with the user's artistic vision. By providing a flexible and intuitive interface, YuE2PlanSong empowers users to experiment with different musical styles and lyrical themes, making it an invaluable tool for AI artists seeking to explore new creative possibilities.

YuE2 生成乐谱计划 Input Parameters:

model

The model parameter specifies the YuE2 model to be used for generating the symbolic plan. This parameter is crucial as it determines the underlying architecture and capabilities of the music generation process. The model choice can significantly impact the style and quality of the generated music, allowing users to tailor the output to their specific needs.

style

The style parameter allows you to input a description of the musical style you wish to emulate. This can include genres, instrumentation, and vocal characteristics. The parameter supports multiline input, enabling detailed and nuanced style descriptions. By defining the style, you guide the node in shaping the musical elements to match your artistic intent.

lyrics

The lyrics parameter is where you input the textual content of the song. Like the style parameter, it supports multiline input, allowing for comprehensive lyrical content. The lyrics serve as a narrative or thematic guide for the music generation process, influencing the emotional and structural aspects of the resulting composition.

cot

The cot parameter stands for "condition on tokens" and offers options such as "full" and "melody." This setting determines whether the node should generate a complete musical plan with melody and chords or focus solely on the melody. The default value is "full," which is suitable for creating original compositions, while "melody" is recommended for covers.

seed

The seed parameter is an integer value used to initialize the random number generator for the music generation process. It ensures reproducibility of results, allowing you to generate the same musical output across different sessions. The default value is 831001, with a minimum of 0 and a maximum of 0x7fffffffffffffff, providing a wide range of possible variations.

YuE2 生成乐谱计划 Output Parameters:

symbolic_plan

The symbolic_plan output provides a structured representation of the musical plan generated by the node. This plan includes encoded musical elements such as melody, harmony, and rhythm, which can be further processed to synthesize audio. The symbolic plan is essential for understanding the composition's structure and serves as a blueprint for subsequent stages in the music generation pipeline.

timing_info

The timing_info output contains detailed information about the timing and duration of the generated musical elements. This data is crucial for synchronizing different parts of the composition and ensuring a coherent musical flow. It provides insights into the temporal structure of the music, aiding in the refinement and adjustment of the final output.

YuE2 生成乐谱计划 Usage Tips:

• Experiment with different style and lyrics combinations to explore a wide range of musical possibilities and discover unique compositions.
• Use the seed parameter to reproduce specific musical outputs or to explore variations by changing the seed value.
• Select the cot option that best suits your project needs; use "full" for original compositions and "melody" for cover versions.

YuE2 生成乐谱计划 Common Errors and Solutions:

"连接的 YuE2 训练节点只完成了预处理，尚未产生歌曲风格模型"

• Explanation: This error indicates that the connected YuE2 training node has only completed preprocessing and has not yet produced a song style model.
• Solution: Ensure that the training process is fully completed and that a valid song style model is available before attempting to generate a symbolic plan.

"训练得到的歌曲风格 LoRA 当前只支持直接生成（cot=off）"

• Explanation: This error occurs when the trained song style LoRA only supports direct generation, but the cot parameter is not set to "off."
• Solution: Set the cot parameter to "off" to align with the capabilities of the trained song style LoRA and proceed with the generation process.