ComfyUI Node: ACE-Step Repaint

ACE-Step Repaint:

The ACE_STEP_Repaint node is designed to modify specific regions of an audio signal by applying a style transformation based on a given description. This node is particularly useful for artists and creators who wish to enhance or alter parts of an audio track to match a desired aesthetic or thematic style. By leveraging advanced model configurations, ACE_STEP_Repaint allows for precise control over the start and end times of the repainting process, ensuring that only the intended sections of the audio are affected. This capability is essential for creating dynamic and engaging audio experiences, as it enables the seamless integration of stylistic changes without disrupting the overall flow of the audio piece.

ACE-Step Repaint Input Parameters:

src_audio

The src_audio parameter represents the original audio signal that you wish to repaint. This input is crucial as it serves as the base material for the transformation process. The node will apply the specified style changes to this audio, ensuring that the output retains the original's core characteristics while incorporating the new stylistic elements.

caption

The caption parameter is a string input that provides a style description prompt for the repainted region. This description guides the transformation process, allowing you to specify the desired aesthetic or thematic changes. The default value is an empty string, and it supports multiline input, enabling detailed and nuanced style descriptions.

repainting_start

The repainting_start parameter is a float that determines the start time for the repainting region in seconds. It allows you to specify precisely where the transformation should begin within the audio track. The default value is 0.0, with a minimum of 0.0 and a maximum of 600.0 seconds, providing flexibility in selecting the repainting region.

repainting_end

The repainting_end parameter is a float that sets the end time for the repainting region in seconds. A value of -1.0 indicates that the repainting should continue until the end of the audio. This parameter allows for precise control over the duration of the transformation, with a default of -1.0 and a range from -1.0 to 600.0 seconds.

checkpoint_dir

The checkpoint_dir parameter specifies the directory containing the ACE-Step model weights, particularly the DiT model. This input is essential for accessing the necessary model configurations required for the repainting process. The default value is the first directory returned by the get_acestep_checkpoints() function.

config_path

The config_path parameter indicates the specific model configuration to use, such as acestep-v15-turbo. This input determines the model's behavior and capabilities during the repainting process, allowing you to select the most suitable configuration for your needs. The default value is acestep-v15-turbo.

lm_model_path

The lm_model_path parameter provides the path to the language model used for processing metadata. This input is crucial for ensuring that the transformation process accurately interprets and applies the style description. The default value is acestep-5Hz-lm-1.7B.

seed

The seed parameter is an integer that sets the random seed for reproducibility. By specifying a seed value, you can ensure consistent results across multiple runs. A value of -1 indicates random generation, while the range extends from -1 to 0xFFFFFFFFFFFFFFFF, with a default of -1.

inference_steps

The inference_steps parameter is an integer that determines the number of diffusion steps used during the repainting process. Higher values, such as 25-50, can improve the quality of the transformation but may result in slower processing times. The default value is 8, with a range from 1 to 64.

ACE-Step Repaint Output Parameters:

repainted_audio

The repainted_audio parameter represents the output audio signal after the repainting process has been applied. This output retains the original audio's core characteristics while incorporating the specified stylistic changes, resulting in a transformed audio experience that aligns with the provided style description.

ACE-Step Repaint Usage Tips:

• To achieve the best results, provide a detailed and specific style description in the caption parameter to guide the transformation process effectively.
• Experiment with different inference_steps values to balance between processing time and output quality, especially for complex transformations.
• Use the seed parameter to ensure reproducibility when you need consistent results across multiple runs.

ACE-Step Repaint Common Errors and Solutions:

"Invalid audio input"

• Explanation: This error occurs when the src_audio parameter does not contain a valid audio signal.
• Solution: Ensure that the input audio file is correctly formatted and accessible by the node.

"Model configuration not found"

• Explanation: This error indicates that the specified config_path does not match any available model configurations.
• Solution: Verify that the config_path is correctly set to an existing model configuration, such as acestep-v15-turbo.

"Repainting region out of bounds"

• Explanation: This error arises when the repainting_start or repainting_end values exceed the audio's duration.
• Solution: Adjust the repainting_start and repainting_end parameters to ensure they fall within the audio's length.