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

fix: index_root searching

close #1147
This commit is contained in:
源文雨 2023-09-01 14:11:55 +08:00
parent d634c2727e
commit 8ffdcb0128
21 changed files with 59 additions and 76 deletions

2
.env
View File

@ -4,5 +4,5 @@ no_proxy = localhost, 127.0.0.1, ::1
# You can change the location of the model, etc. by changing here
weight_root = assets/weights
weight_uvr5_root = assets/uvr5_weights
index_root = output
index_root = logs
rmvpe_root = assets/rmvpe

View File

@ -160,7 +160,7 @@ class Config:
x_center = 30
x_max = 32
if self.dml:
print("use DirectML instead")
print("Use DirectML instead")
if (
os.path.exists(
"runtime\Lib\site-packages\onnxruntime\capi\DirectML.dll"
@ -188,7 +188,7 @@ class Config:
self.is_half = False
else:
if self.instead:
print(f"use {self.instead} instead")
print(f"Use {self.instead} instead")
if (
os.path.exists(
"runtime\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll"

View File

@ -356,7 +356,7 @@ if __name__ == "__main__":
)
if event == "start_vc" and self.flag_vc == False:
if self.set_values(values) == True:
print("using_cuda:" + str(torch.cuda.is_available()))
print("Use CUDA:" + str(torch.cuda.is_available()))
self.start_vc()
settings = {
"pth_path": values["pth_path"],
@ -623,7 +623,7 @@ if __name__ == "__main__":
sola_offset = sola_offset.item()
else:
sola_offset = torch.argmax(cor_nom[0, 0] / cor_den[0, 0])
print("sola offset: " + str(int(sola_offset)))
print("sola_offset =" + str(int(sola_offset)))
self.output_wav[:] = infer_wav[sola_offset : sola_offset + self.block_frame]
self.output_wav[: self.crossfade_frame] *= self.fade_in_window
self.output_wav[: self.crossfade_frame] += self.sola_buffer[:]
@ -663,7 +663,7 @@ if __name__ == "__main__":
outdata[:] = self.output_wav[:].repeat(2, 1).t().cpu().numpy()
total_time = time.perf_counter() - start_time
self.window["infer_time"].update(int(total_time * 1000))
print("infer time:" + str(total_time))
print("Infer time:" + str(total_time))
def get_devices(self, update: bool = True):
"""获取设备列表"""
@ -716,9 +716,9 @@ if __name__ == "__main__":
sd.default.device[1] = output_device_indices[
output_devices.index(output_device)
]
print("input device:" + str(sd.default.device[0]) + ":" + str(input_device))
print("Input device:" + str(sd.default.device[0]) + ":" + str(input_device))
print(
"output device:" + str(sd.default.device[1]) + ":" + str(output_device)
"Output device:" + str(sd.default.device[1]) + ":" + str(output_device)
)
gui = GUI()

View File

@ -124,7 +124,7 @@ class ToolButton(gr.Button, gr.components.FormComponent):
weight_root = os.getenv("weight_root")
weight_uvr5_root = os.getenv("weight_uvr5_root")
index_root = "logs"
index_root = os.getenv("index_root")
names = []
for name in os.listdir(weight_root):
@ -556,14 +556,14 @@ def click_train(
shuffle(opt)
with open("%s/filelist.txt" % exp_dir, "w") as f:
f.write("\n".join(opt))
print("write filelist done")
print("Write filelist done")
# 生成config#无需生成config
# cmd = python_cmd + " train_nsf_sim_cache_sid_load_pretrain.py -e mi-test -sr 40k -f0 1 -bs 4 -g 0 -te 10 -se 5 -pg pretrained/f0G40k.pth -pd pretrained/f0D40k.pth -l 1 -c 0"
print("use gpus:", gpus16)
print("Use gpus:", gpus16)
if pretrained_G14 == "":
print("no pretrained Generator")
print("No pretrained Generator")
if pretrained_D15 == "":
print("no pretrained Discriminator")
print("No pretrained Discriminator")
if gpus16:
cmd = get_quoted_python_cmd() + ' infer/modules/train/train.py -e "%s" -sr %s -f0 %s -bs %s -g %s -te %s -se %s %s %s -l %s -c %s -sw %s -v %s' % (
exp_dir1,

View File

@ -6,7 +6,7 @@ import soundfile
class ContentVec:
def __init__(self, vec_path="pretrained/vec-768-layer-12.onnx", device=None):
print("load model(s) from {}".format(vec_path))
print("Load model(s) from {}".format(vec_path))
if device == "cpu" or device is None:
providers = ["CPUExecutionProvider"]
elif device == "cuda":

View File

@ -51,9 +51,9 @@ def spectrogram_torch(y, n_fft, sampling_rate, hop_size, win_size, center=False)
"""
# Validation
if torch.min(y) < -1.07:
print("min value is ", torch.min(y))
print("spectrogram_torch min value is ", torch.min(y))
if torch.max(y) > 1.07:
print("max value is ", torch.max(y))
print("spectrogram_torch max value is ", torch.max(y))
# Window - Cache if needed
global hann_window

View File

@ -34,7 +34,7 @@ def load_checkpoint_d(checkpoint_path, combd, sbd, optimizer=None, load_opt=1):
new_state_dict[k] = saved_state_dict[k]
if saved_state_dict[k].shape != state_dict[k].shape:
print(
"shape-%s-mismatch|need-%s|get-%s"
"shape-%s-mismatch. need: %s, get: %s"
% (k, state_dict[k].shape, saved_state_dict[k].shape)
) #
raise KeyError

View File

@ -85,7 +85,7 @@ class FeatureInput(object):
if hasattr(self, "model_rmvpe") == False:
from infer.lib.rmvpe import RMVPE
print("loading rmvpe model")
print("Loading rmvpe model")
self.model_rmvpe = RMVPE(
"assets/rmvpe/rmvpe.pt", is_half=False, device="cpu"
)

View File

@ -48,7 +48,7 @@ class FeatureInput(object):
if hasattr(self, "model_rmvpe") == False:
from infer.lib.rmvpe import RMVPE
print("loading rmvpe model")
print("Loading rmvpe model")
self.model_rmvpe = RMVPE(
"assets/rmvpe/rmvpe.pt", is_half=is_half, device="cuda"
)

View File

@ -46,7 +46,7 @@ class FeatureInput(object):
if hasattr(self, "model_rmvpe") == False:
from infer.lib.rmvpe import RMVPE
print("loading rmvpe model")
print("Loading rmvpe model")
self.model_rmvpe = RMVPE(
"assets/rmvpe/rmvpe.pt", is_half=False, device=device
)

View File

@ -101,7 +101,7 @@ class Predictor:
"CPUExecutionProvider",
],
)
print("onnx load done")
print("ONNX load done")
def demix(self, mix):
samples = mix.shape[-1]

View File

@ -90,7 +90,7 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
del pre_fun
except:
traceback.print_exc()
print("clean_empty_cache")
if torch.cuda.is_available():
torch.cuda.empty_cache()
print("Executed torch.cuda.empty_cache()")
yield "\n".join(infos)

View File

@ -31,7 +31,7 @@ class VC:
def get_vc(self, sid, *to_return_protect):
person = f'{os.getenv("weight_root")}/{sid}'
print(f"loading {person}")
print(f"Loading: {person}")
self.cpt = torch.load(person, map_location="cpu")
self.tgt_sr = self.cpt["config"][-1]
@ -77,6 +77,7 @@ class VC:
self.pipeline = Pipeline(self.tgt_sr, self.config)
n_spk = self.cpt["config"][-3]
index = {"value": get_index_path_from_model(sid), "__type__": "update"}
print("Select index:", index["value"])
return (
(

View File

@ -140,7 +140,7 @@ class Pipeline(object):
from infer.lib.rmvpe import RMVPE
print(
"loading rmvpe model,%s" % "%s/rmvpe.pt" % os.environ["rmvpe_root"]
"Loading rmvpe model,%s" % "%s/rmvpe.pt" % os.environ["rmvpe_root"]
)
self.model_rmvpe = RMVPE(
"%s/rmvpe.pt" % os.environ["rmvpe_root"],
@ -152,7 +152,7 @@ class Pipeline(object):
if "privateuseone" in str(self.device): # clean ortruntime memory
del self.model_rmvpe.model
del self.model_rmvpe
print("cleaning ortruntime memory")
print("Cleaning ortruntime memory")
f0 *= pow(2, f0_up_key / 12)
# with open("test.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()]))
@ -262,17 +262,12 @@ class Pipeline(object):
feats = feats.to(feats0.dtype)
p_len = torch.tensor([p_len], device=self.device).long()
with torch.no_grad():
if pitch is not None and pitchf is not None:
audio1 = (
(net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0])
.data.cpu()
.float()
.numpy()
)
else:
audio1 = (
(net_g.infer(feats, p_len, sid)[0][0, 0]).data.cpu().float().numpy()
)
hasp = pitch is not None and pitchf is not None
arg = (feats, p_len, pitch, pitchf, sid) if hasp else (feats, p_len, sid)
audio1 = (
(net_g.infer(*arg)[0][0, 0]).data.cpu().float().numpy()
)
del hasp, arg
del feats, p_len, padding_mask
if torch.cuda.is_available():
torch.cuda.empty_cache()

View File

@ -2,33 +2,20 @@ import os
from fairseq import checkpoint_utils
### don't modify the code before you test it
# def get_index_path_from_model(sid):
# return next(
# (
# f
# for f in [
# os.path.join(root, name)
# for root, dirs, files in os.walk(os.getenv("index_root"), topdown=False)
# for name in files
# if name.endswith(".index") and "trained" not in name
# ]
# if sid.split(".")[0] in f
# ),
# "",
# )
def get_index_path_from_model(sid):
sel_index_path = ""
name = os.path.join("logs", sid.split(".")[0], "")
# print(name)
for f in index_paths:
if name in f:
# print("selected index path:", f)
sel_index_path = f
break
return sel_index_path
return next(
(
f
for f in [
os.path.join(root, name)
for root, _, files in os.walk(os.getenv("index_root"), topdown=False)
for name in files
if name.endswith(".index") and "trained" not in name
]
if sid.split(".")[0] in f
),
"",
)
def load_hubert(config):

View File

@ -23,7 +23,7 @@ vc = VC(config)
weight_root = os.getenv("weight_root")
weight_uvr5_root = os.getenv("weight_uvr5_root")
index_root = "logs"
index_root = os.getenv("index_root")
names = []
hubert_model = None
for name in os.listdir(weight_root):

View File

@ -55,7 +55,7 @@ def main(path, root):
torch.manual_seed(114514)
model_a = torch.load(path, map_location="cpu")["weight"]
print("query:\t\t%s\t%s" % (path, model_hash(path)))
print("Query:\t\t%s\t%s" % (path, model_hash(path)))
map_attn_a = {}
map_rand_input = {}
@ -83,7 +83,7 @@ def main(path, root):
sims.append(sim)
print(
"reference:\t%s\t%s\t%s"
"Reference:\t%s\t%s\t%s"
% (path, model_hash(path), f"{torch.mean(torch.stack(sims)) * 1e2:.2f}%")
)

View File

@ -34,7 +34,7 @@ from scipy.io import wavfile
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_path = r"E:\codes\py39\vits_vc_gpu_train\assets\hubert\hubert_base.pt" #
print("load model(s) from {}".format(model_path))
print("Load model(s) from {}".format(model_path))
models, saved_cfg, task = checkpoint_utils.load_model_ensemble_and_task(
[model_path],
suffix="",

View File

@ -50,14 +50,14 @@ np.save("tools/infer/big_src_feature_mi.npy", big_npy)
# big_npy=np.load("/bili-coeus/jupyter/jupyterhub-liujing04/vits_ch/inference_f0/big_src_feature_mi.npy")
n_ivf = min(int(16 * np.sqrt(big_npy.shape[0])), big_npy.shape[0] // 39)
index = faiss.index_factory(768, "IVF%s,Flat" % n_ivf) # mi
print("training")
print("Training...")
index_ivf = faiss.extract_index_ivf(index) #
index_ivf.nprobe = 1
index.train(big_npy)
faiss.write_index(
index, "tools/infer/trained_IVF%s_Flat_baseline_src_feat_v2.index" % (n_ivf)
)
print("adding")
print("Adding...")
batch_size_add = 8192
for i in range(0, big_npy.shape[0], batch_size_add):
index.add(big_npy[i : i + batch_size_add])

View File

@ -20,12 +20,12 @@ np.save("infer/big_src_feature_mi.npy", big_npy)
# big_npy=np.load("/bili-coeus/jupyter/jupyterhub-liujing04/vits_ch/inference_f0/big_src_feature_mi.npy")
print(big_npy.shape)
index = faiss.index_factory(256, "IVF512,Flat") # mi
print("training")
print("Training...")
index_ivf = faiss.extract_index_ivf(index) #
index_ivf.nprobe = 9
index.train(big_npy)
faiss.write_index(index, "infer/trained_IVF512_Flat_mi_baseline_src_feat.index")
print("adding")
print("Adding...")
index.add(big_npy)
faiss.write_index(index, "infer/added_IVF512_Flat_mi_baseline_src_feat.index")
"""

View File

@ -67,7 +67,7 @@ class RVC:
if index_rate != 0:
self.index = faiss.read_index(index_path)
self.big_npy = self.index.reconstruct_n(0, self.index.ntotal)
print("index search enabled")
print("Index search enabled")
self.index_path = index_path
self.index_rate = index_rate
models, _, _ = fairseq.checkpoint_utils.load_model_ensemble_and_task(
@ -120,7 +120,7 @@ class RVC:
if new_index_rate != 0 and self.index_rate == 0:
self.index = faiss.read_index(self.index_path)
self.big_npy = self.index.reconstruct_n(0, self.index.ntotal)
print("index search enabled")
print("Index search enabled")
self.index_rate = new_index_rate
def get_f0_post(self, f0):
@ -237,7 +237,7 @@ class RVC:
if hasattr(self, "model_rmvpe") == False:
from infer.lib.rmvpe import RMVPE
print("loading rmvpe model")
print("Loading rmvpe model")
self.model_rmvpe = RMVPE(
# "rmvpe.pt", is_half=self.is_half if self.device.type!="privateuseone" else False, device=self.device if self.device.type!="privateuseone"else "cpu"####dml时强制对rmvpe用cpu跑
# "rmvpe.pt", is_half=False, device=self.device####dml配置
@ -295,10 +295,10 @@ class RVC:
+ (1 - self.index_rate) * feats[0][-leng_replace_head:]
)
else:
print("index search FAIL or disabled")
print("Index search FAILED or disabled")
except:
traceback.print_exc()
print("index search FAIL")
print("Index search FAILED")
feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
t3 = ttime()
if self.if_f0 == 1:
@ -338,5 +338,5 @@ class RVC:
.float()
)
t5 = ttime()
print("time->fea-index-f0-model:", t2 - t1, t3 - t2, t4 - t3, t5 - t4)
print("Spent time: fea =", t2 - t1, ", index =", t3 - t2, ", f0 =", t4 - t3, ", model =", t5 - t4)
return infered_audio