Check Vocals:
PlenioVocalCheck is a specialized node designed to analyze audio tracks for the presence of vocals, particularly in instrumental takes. Its primary function is to evaluate multiple audio takes and identify the one with the least vocal interference, ensuring that the best instrumental version is selected. The node utilizes a re-transcription method with SheetSage2 to detect vocal notes within the audio. It prioritizes takes that are free of vocal notes, or if none are entirely free, it selects the one with the least vocal presence. Additionally, it provides insights into takes that may end abruptly while the music is still playing. This node serves as a measurement tool rather than a definitive guarantee, offering valuable feedback for audio production by highlighting potential issues with vocal presence in instrumental tracks.
Check Vocals Input Parameters:
audio
This parameter accepts one or more audio takes, which can be a single take or a list of takes accumulated from a loop. It serves as the primary input for the node to analyze and determine the presence of vocals. The quality and number of takes provided can significantly impact the node's ability to accurately identify the best instrumental version.
audio_encoder
The audio_encoder parameter requires the SheetSage2 encoder, which is essential for the node's operation. This encoder is responsible for processing the audio and identifying vocal notes. Without the correct encoder, the node cannot perform its analysis, making it crucial to ensure that SheetSage2 is loaded and connected properly.
brief
This optional parameter allows you to provide a Song Brief or Cover Brief. It is used to determine whether the takes are instrumental or sung. If the brief indicates a sung song, the node will skip the vocal check and pass the first take unchecked. This parameter helps tailor the node's operation based on the nature of the audio content.
tolerance_seconds
This parameter defines the tolerance level for vocal-like notes that can still be accepted in the analysis. It is a float value with a default setting, allowing you to specify the number of seconds of vocal-like notes that are permissible. The range is from 0.0 to 60.0 seconds, with a step of 0.5. Setting this to 0 means any vocal note will cause the take to fail, while higher values allow for some vocal presence, calibrated against the owner's listening preferences.
Check Vocals Output Parameters:
audio
This output provides the best take, which is the audio version with the least vocal interference. It is the primary result of the node's analysis, offering you the most suitable instrumental track based on the criteria set by the node.
passed
The passed output is a boolean value indicating whether the best take has no vocal notes. A value of True signifies that the selected take is free of vocals, while False indicates the presence of vocal notes in the best take.
report
This output delivers a detailed report containing per-take measurements, the verdict on each take, and any limits encountered during the analysis. It provides a comprehensive overview of the node's findings, helping you understand the vocal presence in each take and the rationale behind the selection of the best take.
takes
The takes output is a list of all the audio takes analyzed, with the best take listed first. This allows you to review all the takes in order of their suitability, providing a clear view of the node's ranking and selection process. You can connect a Preview Audio to listen to each take and verify the node's analysis.
Check Vocals Usage Tips:
- Ensure that the SheetSage2 encoder is properly loaded and connected to the node to enable accurate vocal detection.
- Use the tolerance_seconds parameter to adjust the sensitivity of the node to vocal-like notes, depending on your specific needs for instrumental purity.
- Provide a Song Brief or Cover Brief to optimize the node's performance by indicating whether the takes are instrumental or sung.
Check Vocals Common Errors and Solutions:
"Check Vocals needs the SheetSage2 audio encoder."
- Explanation: This error occurs when the required SheetSage2 encoder is not connected to the node.
- Solution: Load the sheetsage2_bf16.safetensors with the Audio Encoder Loader and ensure it is properly connected to the node.
"sung song: no vocal check"
- Explanation: This message indicates that the node has skipped the vocal check because the provided brief suggests the song is sung.
- Solution: If you intended to check instrumental takes, ensure that the brief correctly reflects the nature of the audio content. If the song is indeed sung, no action is needed as the node is functioning as expected.