OPENGIT/Retrieval-based-Voice-Conversion-WebUI
1
0
Fork 0
mirror of https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI.git synced 2025-01-31 02:32:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

增加无f0模型的支持

Browse Source 
增加无f0模型的支持
This commit is contained in:
RVC-Boss 2023-04-22 11:38:00 +00:00 committed by GitHub
parent 2f51e932bf
commit c423f77a16
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 25 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
gui.py
View File

@ -54,8 +54,8 @@ class RVC:
cpt = torch.load(pth_path, map_location="cpu") cpt = torch.load(pth_path, map_location="cpu")
tgt_sr = cpt["config"][-1] tgt_sr = cpt["config"][-1]
cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] # n_spk cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] # n_spk
if_f0 = cpt.get("f0", 1) self.if_f0 = cpt.get("f0", 1)
if if_f0 == 1: if self.if_f0 == 1:
self.net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=True) self.net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=True)
else: else:
self.net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"]) self.net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
@ -136,27 +136,37 @@ class RVC:
feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1) feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
torch.cuda.synchronize() torch.cuda.synchronize()
# p_len = min(feats.shape[1],10000,pitch.shape[0])#太大了爆显存
p_len = min(feats.shape[1], 12000) #
print(feats.shape) print(feats.shape)
pitch, pitchf = self.get_f0(audio, self.f0_up_key) if(self.if_f0==1):
p_len = min(feats.shape[1], 12000, pitch.shape[0]) # 太大了爆显存 pitch, pitchf = self.get_f0(audio, self.f0_up_key)
p_len = min(feats.shape[1], 13000, pitch.shape[0]) # 太大了爆显存
else:
pitch, pitchf = None, None
p_len = min(feats.shape[1], 13000) # 太大了爆显存
torch.cuda.synchronize() torch.cuda.synchronize()
# print(feats.shape,pitch.shape) # print(feats.shape,pitch.shape)
feats = feats[:, :p_len, :] feats = feats[:, :p_len, :]
pitch = pitch[:p_len] if(self.if_f0==1):
pitchf = pitchf[:p_len] pitch = pitch[:p_len]
pitchf = pitchf[:p_len]
pitch = torch.LongTensor(pitch).unsqueeze(0).to(device)
pitchf = torch.FloatTensor(pitchf).unsqueeze(0).to(device)
p_len = torch.LongTensor([p_len]).to(device) p_len = torch.LongTensor([p_len]).to(device)
pitch = torch.LongTensor(pitch).unsqueeze(0).to(device)
pitchf = torch.FloatTensor(pitchf).unsqueeze(0).to(device)
ii = 0 # sid ii = 0 # sid
sid = torch.LongTensor([ii]).to(device) sid = torch.LongTensor([ii]).to(device)
with torch.no_grad(): with torch.no_grad():
infered_audio = ( if(self.if_f0==1):
self.net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0] infered_audio = (
.data.cpu() self.net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0]
.float() .data.cpu()
) # nsf .float()
)
else:
infered_audio = (
self.net_g.infer(feats, p_len, sid)[0][0, 0]
.data.cpu()
.float()
)
torch.cuda.synchronize() torch.cuda.synchronize()
return infered_audio return infered_audio