mc.nn~ processes multiple audio channels through one RAVE model instance to cut CPU and RAM

In Max 8, mc.nn~ (multi-channel nn~) uses the multi-channel feature of Max 8 to perform batch processing: instead of instantiating multiple nn~ objects that each load a separate copy of the model into memory, mc.nn~ routes several inputs through a single model instance. This leverages the batch dimension of the neural network, which is far more efficient. The result is minimum CPU cost and a single model in RAM. This is relevant for complex performance patches where a performer wants to process several input sources (drums, bass, voice) through the same timbre model simultaneously.

Examples

In Max 8: mc.nn~ my_model forward 2048 with an mc signal bundle of 4 channels → outputs 4 independently transformed audio streams from one loaded model.

Assessment

A performer wants to timbre-transfer four drum microphone channels through the same RAVE model in Max. Compare using four nn~ objects versus one mc.nn~ object in terms of CPU and RAM usage.