VITS-Umamusume-voice-synthesizer / ONNXVITS_to_onnx.py
Plachta's picture
Upload 35 files
35aaf1e
raw
history blame
1.05 kB
import ONNXVITS_models
import utils
from text import text_to_sequence
import torch
import commons
def get_text(text, hps):
text_norm = text_to_sequence(text, hps.symbols, hps.data.text_cleaners)
if hps.data.add_blank:
text_norm = commons.intersperse(text_norm, 0)
text_norm = torch.LongTensor(text_norm)
return text_norm
hps = utils.get_hparams_from_file("../vits/pretrained_models/uma87.json")
symbols = hps.symbols
net_g = ONNXVITS_models.SynthesizerTrn(
len(symbols),
hps.data.filter_length // 2 + 1,
hps.train.segment_size // hps.data.hop_length,
n_speakers=hps.data.n_speakers,
**hps.model)
_ = net_g.eval()
_ = utils.load_checkpoint("../vits/pretrained_models/uma_1153000.pth", net_g)
text1 = get_text("γ‚γ‚ŠγŒγ¨γ†γ”γ–γ„γΎγ™γ€‚", hps)
stn_tst = text1
with torch.no_grad():
x_tst = stn_tst.unsqueeze(0)
x_tst_lengths = torch.LongTensor([stn_tst.size(0)])
sid = torch.tensor([0])
o = net_g(x_tst, x_tst_lengths, sid=sid, noise_scale=.667, noise_scale_w=0.8, length_scale=1)