# Sep

Sep: 0
Sep: <Stem>[; <Model>]

Separate the input audio into multiple stems, or isolate a specific stem from the rest of the song.

<Stem>: The stem to isolate from the rest of the song. Refer to the constants for the allowed values.
Number: more than -3, less than 7
<Model>: The AI model to use to split the input song into stems. Refer to the constants for the allowed values.
String: any model supported by demucs

The <Model> parameter is only available on version 1.2 and later of the Scene specification. Make sure your Scene file is using version 1.2 or later if you want to use the <Model> parameter.
In version 1.3 or later, it is optional if you are using STEMS_MANUAL or STEMS_SIMPLE.
In version 1.4 or later, it is also optional if you are using STEMS_COPY.