Table 1.

The searched algorithm and hyperparameter space.

Neural decoders
Kalman filter	C: 0.1, 1, 10
Wiener filter
Wiener cascade	Degree: 2, 3, 4, 5
Dense Neural Network (NN)	# Units: 256, 512, 1024, 2048
Simple Recurrent NN (RNN)	# Units: 256, 512, 1024, 2048
Gated Recurrent Unit (GRU) RNN	# Units: 256, 512, 1024, 2048
Long Short-Term Memory (LSTM) RNN	# Units: 256, 512, 1024, 2048

Neural feature extraction
Array Location:	rostral parabelt (RPB), caudal parabelt (CPB)
Filter Low Cutoff (Hz):	100 to 1000, increments of 100
Filter High Cutoff (Hz):	1000 to 10000, increments of 1000
Threshold Factor:	2, 3, 4, 5
# Channels:	2, 4, 8, 16, 32, 64, 96
Feature Window Span:	4, 8, 16

Audio processing
Sounds:	“tree”, “good”, “north”, “cricket”, “program”, macaque call
# Sounds:	1, 2, 3, 4, 5, 6
# Mel-Bands:	32, 64, 128, 256
Hop Size (ms):	10, 20, 30, 40, 50