class SpeechEncoder(object): def __init__(self, vec_path="pretrain/checkpoint_best_legacy_500.pt", device=None): self.model = None # This is Model self.hidden_dim = 768 pass def encoder(self, wav): """ input: wav:[signal_length] output: embedding:[batchsize,hidden_dim,wav_frame] """ pass