FreeVC / speaker_encoder /params_model.py
kevinwang676's picture
Duplicate from OlaWod/FreeVC
b84b595
raw
history blame contribute delete
196 Bytes
## Model parameters
model_hidden_size = 256
model_embedding_size = 256
model_num_layers = 3
## Training parameters
learning_rate_init = 1e-4
speakers_per_batch = 64
utterances_per_speaker = 10