Instantiate a hubert feature reader from checkpoint file at `checkpoint_path`, setup to extract features from `layer` of the model. Optionally specify which `device` to perform inference on and the maximum chunk length to process audio on (chunk length is in samples). Note:
(self, checkpoint_path, layer: int, max_chunk=1600000, device='cuda', downsample_factor=320)
source not stored for this graph (policy: none)
nothing calls this directly
no outgoing calls
no test coverage detected