so-vits-svc/vencoder/encoder.py

14 lines
364 B
Python

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