What are interleaved separations?

When separating one clip with multiple models, for example when separating vocals, drums and bass, the order of the separations can be done two different ways.
The default behaviour is to separate the whole clip with a single model before moving to the next model. This is the fastest way to run a separation but this means you will need to wait until the end of the vocal separation before being able to listen to the drums or bass.
The other option is to "interleave" the separations, which will now separate the first section of vocals, followed by the first section of drums and so on, swapping between the models. This can be useful if you want to quickly listen to each of the separated sources during processing. However, it is important to note that this comes at the cost of repeatedly loading and unloading the models and can result in significantly slower separations on cpu and on low performance GPUs.
Interleaved separations can be toggled in the Processing tab in the global settings, by checking the "Run separations interleaved" button.

Get App?