1
0
mirror of synced 2024-11-23 23:21:03 +01:00

Format code (#330)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
This commit is contained in:
github-actions[bot] 2023-05-21 19:19:53 +08:00 committed by GitHub
parent 067731db9b
commit cfd9848128
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 39 additions and 22 deletions

View File

@ -2,6 +2,7 @@ import argparse
import torch
from multiprocessing import cpu_count
def config_file_change_fp32():
for config_file in ["32k.json", "40k.json", "48k.json"]:
with open(f"configs/{config_file}", "r") as f:
@ -13,6 +14,7 @@ def config_file_change_fp32():
with open("trainset_preprocess_pipeline_print.py", "w") as f:
f.write(strr)
class Config:
def __init__(self):
self.device = "cuda:0"

View File

@ -52,4 +52,3 @@ if __name__ == "__main__":
input_names=input_names,
output_names=output_names,
)

44
gui.py
View File

@ -1,4 +1,4 @@
'''
"""
0416后的更新
引入config中half
重建npy而不用填写
@ -9,12 +9,14 @@
int16
增加无索引支持
f0算法改harvest(怎么看就只有这个会影响CPU占用)但是不这么改效果不好
'''
"""
import os, sys, traceback
now_dir = os.getcwd()
sys.path.append(now_dir)
from config import Config
is_half=Config().is_half
is_half = Config().is_half
import PySimpleGUI as sg
import sounddevice as sd
import noisereduce as nr
@ -26,7 +28,12 @@ import torchaudio.transforms as tat
import scipy.signal as signal
# import matplotlib.pyplot as plt
from infer_pack.models import SynthesizerTrnMs256NSFsid, SynthesizerTrnMs256NSFsid_nono,SynthesizerTrnMs768NSFsid,SynthesizerTrnMs768NSFsid_nono
from infer_pack.models import (
SynthesizerTrnMs256NSFsid,
SynthesizerTrnMs256NSFsid_nono,
SynthesizerTrnMs768NSFsid,
SynthesizerTrnMs768NSFsid_nono,
)
from i18n import I18nAuto
i18n = I18nAuto()
@ -63,7 +70,7 @@ class RVC:
)
self.model = models[0]
self.model = self.model.to(device)
if(is_half==True):
if is_half == True:
self.model = self.model.half()
else:
self.model = self.model.float()
@ -75,21 +82,25 @@ class RVC:
self.version = cpt.get("version", "v1")
if version == "v1":
if if_f0 == 1:
self.net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=config.is_half)
self.net_g = SynthesizerTrnMs256NSFsid(
*cpt["config"], is_half=config.is_half
)
else:
self.net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
elif version == "v2":
if if_f0 == 1:
self.net_g = SynthesizerTrnMs768NSFsid(*cpt["config"], is_half=config.is_half)
self.net_g = SynthesizerTrnMs768NSFsid(
*cpt["config"], is_half=config.is_half
)
else:
self.net_g = SynthesizerTrnMs768NSFsid_nono(*cpt["config"])
del self.net_g.enc_q
print(self.net_g.load_state_dict(cpt["weight"], strict=False))
self.net_g.eval().to(device)
if(is_half==True):
self.net_g=self.net_g.half()
if is_half == True:
self.net_g = self.net_g.half()
else:
self.net_g=self.net_g.float()
self.net_g = self.net_g.float()
except:
print(traceback.format_exc())
@ -151,15 +162,18 @@ class RVC:
####索引优化
try:
if hasattr(self, "index") and hasattr(self, "big_npy") and self.index_rate != 0:
if (
hasattr(self, "index")
and hasattr(self, "big_npy")
and self.index_rate != 0
):
npy = feats[0].cpu().numpy().astype("float32")
score, ix = self.index.search(npy, k=8)
weight = np.square(1 / score)
weight /= weight.sum(axis=1, keepdims=True)
npy = np.sum(
self.big_npy[ix] * np.expand_dims(weight, axis=2), axis=1
)
if(is_half==True):npy=npy.astype("float16")
npy = np.sum(self.big_npy[ix] * np.expand_dims(weight, axis=2), axis=1)
if is_half == True:
npy = npy.astype("float16")
feats = (
torch.from_numpy(npy).unsqueeze(0).to(device) * self.index_rate
+ (1 - self.index_rate) * feats

View File

@ -18,11 +18,11 @@ class I18nAuto:
if not os.path.exists(f"./i18n/{language}.json"):
language = "en_US"
self.language = language
#print("Use Language:", language)
# print("Use Language:", language)
self.language_map = load_language_list(language)
def __call__(self, key):
return self.language_map.get(key, key)
def print(self):
print("Use Language:", self.language)

View File

@ -1069,6 +1069,8 @@ def change_info_(ckpt_path):
from infer_pack.models_onnx import SynthesizerTrnMsNSFsidM
def export_onnx(ModelPath, ExportedPath, MoeVS=True):
cpt = torch.load(ModelPath, map_location="cpu")
cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] # n_spk

View File

@ -633,14 +633,14 @@ class SynthesizerTrnMsNSFsidM(nn.Module):
self.speaker_map = self.speaker_map.unsqueeze(0)
def forward(self, phone, phone_lengths, pitch, nsff0, g, rnd, max_len=None):
if self.speaker_map is not None: # [N, S] * [S, B, 1, H]
if self.speaker_map is not None: # [N, S] * [S, B, 1, H]
g = g.reshape((g.shape[0], g.shape[1], 1, 1, 1)) # [N, S, B, 1, 1]
g = g * self.speaker_map # [N, S, B, 1, H]
g = torch.sum(g, dim=1) # [N, 1, B, 1, H]
g = g.transpose(0, -1).transpose(0, -2).squeeze(0) # [B, H, N]
g = torch.sum(g, dim=1) # [N, 1, B, 1, H]
g = g.transpose(0, -1).transpose(0, -2).squeeze(0) # [B, H, N]
else:
g = g.unsqueeze(0)
g = self.emb_g(g).transpose(1,2)
g = self.emb_g(g).transpose(1, 2)
m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
z_p = (m_p + torch.exp(logs_p) * rnd) * x_mask