Mirrors/Mangio-RVC-Fork
1
0
Fork 0
mirror of https://github.com/Mangio621/Mangio-RVC-Fork.git synced 2025-12-16 11:37:44 +01:00
Code Issues Packages Projects Releases Wiki Activity

Revert "Merge branch 'main' into dev-main-backup"

Browse Source 
This reverts commit 81fd2a0528, reversing
changes made to 521f5fb78e.
This commit is contained in:
kalomaze
2023-07-26 22:34:48 -05:00
parent 81fd2a0528
commit 5d25c72bda
8 changed files with 233 additions and 270 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
README.md
View File

@@ -1,5 +1,10 @@
# 7/25 Changelog:
- Better MacOS installation script. For inference, all that needs to be done is running the `run.sh` from the extracted zip folder, where it will install Python 3.8, Homebrew, and other dependencies for you automatically. M1 Macs are natively supported for GPU acceleration, and training should work if you choose to download the pretrained models.
- Fixed the cli inferencing traceback.
- SQL Databases implemented for formanting training stop.
- Gradio browser tab renamed to `Mangio-RVC-Web 💻`.
- Rudimentary functions from `infer-web.py` removed.
- Formanting now accepts any audio format, as long as it is supported by FFmpeg.
# 7/23 Changelog:
- Fp16 detection now works how it did before the last RVC beta; no more training slowdowns compared to how it was before then

6
extract_f0_print.py
View File

@@ -19,10 +19,8 @@ f = open("%s/extract_f0_feature.log" % exp_dir, "a+")
DoFormant = False
with open("formanting.txt", "r") as fvf:
content = fvf.readlines()
Quefrency, Timbre = content[1].split("\n")[0], content[2].split("\n")[0]
Quefrency = 0.0
Timbre = 0.0
def printt(strr):
print(strr)

3
formanting.txt
View File

@@ -1,3 +0,0 @@
False
8.0
1.2

324
infer-web.py
View File

@@ -43,8 +43,44 @@ from train.process_ckpt import change_info, extract_small_model, merge, show_inf
from vc_infer_pipeline import VC
from sklearn.cluster import MiniBatchKMeans
import sqlite3
def clear_sql(signal, frame):
cursor.execute("DELETE FROM formant_data")
cursor.execute("DELETE FROM stop_train")
conn.commit()
conn.close()
print("Clearing SQL database...")
sys.exit(0)
if sys.platform == 'win32':
signal.signal(signal.SIGBREAK, clear_sql)
signal.signal(signal.SIGINT, clear_sql)
signal.signal(signal.SIGTERM, clear_sql)
logging.getLogger("numba").setLevel(logging.WARNING)
conn = sqlite3.connect('TEMP/db:cachedb?mode=memory&cache=shared', check_same_thread=False)
cursor = conn.cursor()
cursor.execute("""
CREATE TABLE IF NOT EXISTS formant_data (
Quefrency FLOAT,
Timbre FLOAT,
DoFormant INTEGER
)
""")
cursor.execute("""
CREATE TABLE IF NOT EXISTS stop_train (
stop BOOL
)
""")
tmp = os.path.join(now_dir, "TEMP")
shutil.rmtree(tmp, ignore_errors=True)
@@ -52,21 +88,29 @@ shutil.rmtree("%s/runtime/Lib/site-packages/infer_pack" % (now_dir), ignore_erro
shutil.rmtree("%s/runtime/Lib/site-packages/uvr5_pack" % (now_dir), ignore_errors=True)
os.makedirs(tmp, exist_ok=True)
os.makedirs(os.path.join(now_dir, "logs"), exist_ok=True)
os.makedirs(os.path.join(now_dir, "audios"), exist_ok=True)
os.makedirs(os.path.join(now_dir, "datasets"), exist_ok=True)
os.makedirs(os.path.join(now_dir, "weights"), exist_ok=True)
os.environ["TEMP"] = tmp
warnings.filterwarnings("ignore")
torch.manual_seed(114514)
DoFormant = False
Quefrency = 8.0
Timbre = 1.2
global DoFormant, Quefrency, Timbre
with open('formanting.txt', 'w+') as fsf:
fsf.truncate(0)
fsf.writelines([str(DoFormant) + '\n', str(Quefrency) + '\n', str(Timbre) + '\n'])
try:
cursor.execute("SELECT Quefrency, Timbre, DoFormant FROM formant_data")
Quefrency, Timbre, DoFormant = cursor.fetchone()
except Exception:
Quefrency = 8.0
Timbre = 1.2
DoFormant = False
cursor.execute("DELETE FROM formant_data")
cursor.execute("DELETE FROM stop_train")
cursor.execute("INSERT INTO formant_data (Quefrency, Timbre, DoFormant) VALUES (?, ?, ?)", (Quefrency, Timbre, 0))
conn.commit()
config = Config()
i18n = I18nAuto()
i18n.print()
@@ -124,20 +168,10 @@ else:
default_batch_size = 1
gpus = "-".join([i[0] for i in gpu_infos])
class ToolButton(gr.Button, gr.components.FormComponent):
"""Small button with single emoji as text, fits inside gradio forms"""
def __init__(self, **kwargs):
super().__init__(variant="tool", **kwargs)
def get_block_name(self):
return "button"
hubert_model = None
def load_hubert():
global hubert_model
models, _, _ = checkpoint_utils.load_model_ensemble_and_task(
@@ -156,10 +190,7 @@ def load_hubert():
weight_root = "weights"
weight_uvr5_root = "uvr5_weights"
index_root = "./logs/"
global audio_root
audio_root = "audios"
global input_audio_path0
global input_audio_path1
names = []
for name in os.listdir(weight_root):
if name.endswith(".pth"):
@@ -228,17 +259,6 @@ def get_fshift_presets():
return ''
def get_audios():
if check_for_name() != '':
audios_path= '"' + os.path.abspath(os.getcwd()) + '/audios/'
if os.path.exists(audios_path):
for file in os.listdir(audios_path):
print(audios_path.join(file) + '"')
return os.path.join(audios_path, file + '"')
return ''
else:
return ''
def vc_single(
sid,
@@ -627,14 +647,13 @@ def if_done_multi(done, ps):
done[0] = True
def formant_enabled(cbox, qfrency, tmbre, frmntapply, formantpreset, formant_refresh_button):
if (cbox):
DoFormant = True
with open('formanting.txt', 'w') as fxxf:
fxxf.truncate(0)
fxxf.writelines([str(DoFormant) + '\n', str(Quefrency) + '\n', str(Timbre) + '\n'])
cursor.execute("DELETE FROM formant_data")
cursor.execute("INSERT INTO formant_data (Quefrency, Timbre, DoFormant) VALUES (?, ?, ?)", (qfrency, tmbre, 1))
conn.commit()
#print(f"is checked? - {cbox}\ngot {DoFormant}")
return (
@@ -650,10 +669,10 @@ def formant_enabled(cbox, qfrency, tmbre, frmntapply, formantpreset, formant_ref
else:
DoFormant = False
with open('formanting.txt', 'w') as fxf:
fxf.truncate(0)
fxf.writelines([str(DoFormant) + '\n', str(Quefrency) + '\n', str(Timbre) + '\n'])
cursor.execute("DELETE FROM formant_data")
cursor.execute("INSERT INTO formant_data (Quefrency, Timbre, DoFormant) VALUES (?, ?, ?)", (qfrency, tmbre, int(DoFormant)))
conn.commit()
#print(f"is checked? - {cbox}\ngot {DoFormant}")
return (
{"value": False, "__type__": "update"},
@@ -670,11 +689,10 @@ def formant_apply(qfrency, tmbre):
Quefrency = qfrency
Timbre = tmbre
DoFormant = True
with open('formanting.txt', 'w') as fxxxf:
fxxxf.truncate(0)
cursor.execute("DELETE FROM formant_data")
cursor.execute("INSERT INTO formant_data (Quefrency, Timbre, DoFormant) VALUES (?, ?, ?)", (qfrency, tmbre, 1))
conn.commit()
fxxxf.writelines([str(DoFormant) + '\n', str(Quefrency) + '\n', str(Timbre) + '\n'])
return ({"value": Quefrency, "__type__": "update"}, {"value": Timbre, "__type__": "update"})
def update_fshift_presets(preset, qfrency, tmbre):
@@ -829,12 +847,12 @@ def change_sr2(sr2, if_f0_3, version19):
if not if_pretrained_generator_exist:
print(
"pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2),
"not exist, will not use pretrained model",
"doesn't exist, will not use pretrained model",
)
if not if_pretrained_discriminator_exist:
print(
"pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2),
"not exist, will not use pretrained model",
"doesn't exist, will not use pretrained model",
)
return (
"pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2)
@@ -865,12 +883,12 @@ def change_version19(sr2, if_f0_3, version19):
if not if_pretrained_generator_exist:
print(
"pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2),
"not exist, will not use pretrained model",
"doesn't exist, will not use pretrained model",
)
if not if_pretrained_discriminator_exist:
print(
"pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2),
"not exist, will not use pretrained model",
"doesn't exist, will not use pretrained model",
)
return (
"pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2)
@@ -972,6 +990,8 @@ def click_train(
if_save_every_weights18,
version19,
):
cursor.execute("DELETE FROM stop_train")
conn.commit()
# 生成filelist
exp_dir = "%s/logs/%s" % (now_dir, exp_dir1)
os.makedirs(exp_dir, exist_ok=True)
@@ -1547,10 +1567,26 @@ def cli_infer(com):
mix = float(com[10])
feature_ratio = float(com[11])
protection_amnt = float(com[12])
#####
protect1 = 0.5
if com[14] == 'False' or com[14] == 'false':
DoFormant = False
Quefrency = 0.0
Timbre = 0.0
cursor.execute("DELETE FROM formant_data")
cursor.execute("INSERT INTO formant_data (Quefrency, Timbre, DoFormant) VALUES (?, ?, ?)", (Quefrency, Timbre, 0))
conn.commit()
else:
DoFormant = True
Quefrency = float(com[15])
Timbre = float(com[16])
cursor.execute("DELETE FROM formant_data")
cursor.execute("INSERT INTO formant_data (Quefrency, Timbre, DoFormant) VALUES (?, ?, ?)", (Quefrency, Timbre, 1))
conn.commit()
print("Mangio-RVC-Fork Infer-CLI: Starting the inference...")
vc_data = get_vc(model_name)
vc_data = get_vc(model_name, protection_amnt, protect1)
print(vc_data)
print("Mangio-RVC-Fork Infer-CLI: Performing inference...")
conversion_data = vc_single(
@@ -1697,7 +1733,6 @@ def preset_apply(preset, qfer, tmbr):
with open(str(preset), 'r') as p:
content = p.readlines()
qfer, tmbr = content[0].split('\n')[0], content[1]
formant_apply(qfer, tmbr)
else:
pass
@@ -1705,70 +1740,86 @@ def preset_apply(preset, qfer, tmbr):
def print_page_details():
if cli_current_page == "HOME":
print(" go home : Takes you back to home with a navigation list.")
print(" go infer : Takes you to inference command execution.\n")
print(" go pre-process : Takes you to training step.1) pre-process command execution.")
print(" go extract-feature : Takes you to training step.2) extract-feature command execution.")
print(" go train : Takes you to training step.3) being or continue training command execution.")
print(" go train-feature : Takes you to the train feature index command execution.\n")
print(" go extract-model : Takes you to the extract small model command execution.")
print(
"\n go home : Takes you back to home with a navigation list."
"\n go infer : Takes you to inference command execution."
"\n go pre-process : Takes you to training step.1) pre-process command execution."
"\n go extract-feature : Takes you to training step.2) extract-feature command execution."
"\n go train : Takes you to training step.3) being or continue training command execution."
"\n go train-feature : Takes you to the train feature index command execution."
"\n go extract-model : Takes you to the extract small model command execution."
)
elif cli_current_page == "INFER":
print(" arg 1) model name with .pth in ./weights: mi-test.pth")
print(" arg 2) source audio path: myFolder\\MySource.wav")
print(" arg 3) output file name to be placed in './audio-outputs': MyTest.wav")
print(" arg 4) feature index file path: logs/mi-test/added_IVF3042_Flat_nprobe_1.index")
print(" arg 5) speaker id: 0")
print(" arg 6) transposition: 0")
print(" arg 7) f0 method: harvest (pm, harvest, crepe, crepe-tiny, hybrid[x,x,x,x], mangio-crepe, mangio-crepe-tiny)")
print(" arg 8) crepe hop length: 160")
print(" arg 9) harvest median filter radius: 3 (0-7)")
print(" arg 10) post resample rate: 0")
print(" arg 11) mix volume envelope: 1")
print(" arg 12) feature index ratio: 0.78 (0-1)")
print(" arg 13) Voiceless Consonant Protection (Less Artifact): 0.33 (Smaller number = more protection. 0.50 means Dont Use.) \n")
print("Example: mi-test.pth saudio/Sidney.wav myTest.wav logs/mi-test/added_index.index 0 -2 harvest 160 3 0 1 0.95 0.33")
print(
"\n arg 1) model name with .pth in ./weights: mi-test.pth"
"\n arg 2) source audio path: myFolder\\MySource.wav"
"\n arg 3) output file name to be placed in './audio-outputs': MyTest.wav"
"\n arg 4) feature index file path: logs/mi-test/added_IVF3042_Flat_nprobe_1.index"
"\n arg 5) speaker id: 0"
"\n arg 6) transposition: 0"
"\n arg 7) f0 method: harvest (pm, harvest, crepe, crepe-tiny, hybrid[x,x,x,x], mangio-crepe, mangio-crepe-tiny, rmvpe)"
"\n arg 8) crepe hop length: 160"
"\n arg 9) harvest median filter radius: 3 (0-7)"
"\n arg 10) post resample rate: 0"
"\n arg 11) mix volume envelope: 1"
"\n arg 12) feature index ratio: 0.78 (0-1)"
"\n arg 13) Voiceless Consonant Protection (Less Artifact): 0.33 (Smaller number = more protection. 0.50 means Dont Use.)"
"\n arg 14) Whether to formant shift the inference audio before conversion: False (if set to false, you can ignore setting the quefrency and timbre values for formanting)"
"\n arg 15)* Quefrency for formanting: 8.0 (no need to set if arg14 is False/false)"
"\n arg 16)* Timbre for formanting: 1.2 (no need to set if arg14 is False/false) \n"
"\nExample: mi-test.pth saudio/Sidney.wav myTest.wav logs/mi-test/added_index.index 0 -2 harvest 160 3 0 1 0.95 0.33 0.45 True 8.0 1.2"
)
elif cli_current_page == "PRE-PROCESS":
print(" arg 1) Model folder name in ./logs: mi-test")
print(" arg 2) Trainset directory: mydataset (or) E:\\my-data-set")
print(" arg 3) Sample rate: 40k (32k, 40k, 48k)")
print(" arg 4) Number of CPU threads to use: 8 \n")
print("Example: mi-test mydataset 40k 24")
print(
"\n arg 1) Model folder name in ./logs: mi-test"
"\n arg 2) Trainset directory: mydataset (or) E:\\my-data-set"
"\n arg 3) Sample rate: 40k (32k, 40k, 48k)"
"\n arg 4) Number of CPU threads to use: 8 \n"
"\nExample: mi-test mydataset 40k 24"
)
elif cli_current_page == "EXTRACT-FEATURE":
print(" arg 1) Model folder name in ./logs: mi-test")
print(" arg 2) Gpu card slot: 0 (0-1-2 if using 3 GPUs)")
print(" arg 3) Number of CPU threads to use: 8")
print(" arg 4) Has Pitch Guidance?: 1 (0 for no, 1 for yes)")
print(" arg 5) f0 Method: harvest (pm, harvest, dio, crepe)")
print(" arg 6) Crepe hop length: 128")
print(" arg 7) Version for pre-trained models: v2 (use either v1 or v2)\n")
print("Example: mi-test 0 24 1 harvest 128 v2")
print(
"\n arg 1) Model folder name in ./logs: mi-test"
"\n arg 2) Gpu card slot: 0 (0-1-2 if using 3 GPUs)"
"\n arg 3) Number of CPU threads to use: 8"
"\n arg 4) Has Pitch Guidance?: 1 (0 for no, 1 for yes)"
"\n arg 5) f0 Method: harvest (pm, harvest, dio, crepe)"
"\n arg 6) Crepe hop length: 128"
"\n arg 7) Version for pre-trained models: v2 (use either v1 or v2)\n"
"\nExample: mi-test 0 24 1 harvest 128 v2"
)
elif cli_current_page == "TRAIN":
print(" arg 1) Model folder name in ./logs: mi-test")
print(" arg 2) Sample rate: 40k (32k, 40k, 48k)")
print(" arg 3) Has Pitch Guidance?: 1 (0 for no, 1 for yes)")
print(" arg 4) speaker id: 0")
print(" arg 5) Save epoch iteration: 50")
print(" arg 6) Total epochs: 10000")
print(" arg 7) Batch size: 8")
print(" arg 8) Gpu card slot: 0 (0-1-2 if using 3 GPUs)")
print(" arg 9) Save only the latest checkpoint: 0 (0 for no, 1 for yes)")
print(" arg 10) Whether to cache training set to vram: 0 (0 for no, 1 for yes)")
print(" arg 11) Save extracted small model every generation?: 0 (0 for no, 1 for yes)")
print(" arg 12) Model architecture version: v2 (use either v1 or v2)\n")
print("Example: mi-test 40k 1 0 50 10000 8 0 0 0 0 v2")
print(
"\n arg 1) Model folder name in ./logs: mi-test"
"\n arg 2) Sample rate: 40k (32k, 40k, 48k)"
"\n arg 3) Has Pitch Guidance?: 1 (0 for no, 1 for yes)"
"\n arg 4) speaker id: 0"
"\n arg 5) Save epoch iteration: 50"
"\n arg 6) Total epochs: 10000"
"\n arg 7) Batch size: 8"
"\n arg 8) Gpu card slot: 0 (0-1-2 if using 3 GPUs)"
"\n arg 9) Save only the latest checkpoint: 0 (0 for no, 1 for yes)"
"\n arg 10) Whether to cache training set to vram: 0 (0 for no, 1 for yes)"
"\n arg 11) Save extracted small model every generation?: 0 (0 for no, 1 for yes)"
"\n arg 12) Model architecture version: v2 (use either v1 or v2)\n"
"\nExample: mi-test 40k 1 0 50 10000 8 0 0 0 0 v2"
)
elif cli_current_page == "TRAIN-FEATURE":
print(" arg 1) Model folder name in ./logs: mi-test")
print(" arg 2) Model architecture version: v2 (use either v1 or v2)\n")
print("Example: mi-test v2")
print(
"\n arg 1) Model folder name in ./logs: mi-test"
"\n arg 2) Model architecture version: v2 (use either v1 or v2)\n"
"\nExample: mi-test v2"
)
elif cli_current_page == "EXTRACT-MODEL":
print(" arg 1) Model Path: logs/mi-test/G_168000.pth")
print(" arg 2) Model save name: MyModel")
print(" arg 3) Sample rate: 40k (32k, 40k, 48k)")
print(" arg 4) Has Pitch Guidance?: 1 (0 for no, 1 for yes)")
print(' arg 5) Model information: "My Model"')
print(" arg 6) Model architecture version: v2 (use either v1 or v2)\n")
print('Example: logs/mi-test/G_168000.pth MyModel 40k 1 "Created by Cole Mangio" v2')
print("")
print(
"\n arg 1) Model Path: logs/mi-test/G_168000.pth"
"\n arg 2) Model save name: MyModel"
"\n arg 3) Sample rate: 40k (32k, 40k, 48k)"
"\n arg 4) Has Pitch Guidance?: 1 (0 for no, 1 for yes)"
'\n arg 5) Model information: "My Model"'
"\n arg 6) Model architecture version: v2 (use either v1 or v2)\n"
'\nExample: logs/mi-test/G_168000.pth MyModel 40k 1 "Created by Cole Mangio" v2'
)
def change_page(page):
global cli_current_page
@@ -1810,7 +1861,7 @@ def execute_command(com):
def cli_navigation_loop():
while True:
print("You are currently in '%s':" % cli_current_page)
print("\nYou are currently in '%s':" % cli_current_page)
print_page_details()
command = input("%s: " % cli_current_page)
try:
@@ -1855,11 +1906,11 @@ def match_index(sid0):
if filename.endswith(".index"):
for i in range(len(indexes_list)):
if indexes_list[i] == (os.path.join(("./logs/" + folder), filename).replace('\\','/')):
print('regular index found')
#print('regular index found')
break
else:
if indexes_list[i] == (os.path.join(("./logs/" + folder.lower()), filename).replace('\\','/')):
print('lowered index found')
#print('lowered index found')
parent_dir = "./logs/" + folder.lower()
break
#elif (indexes_list[i]).casefold() == ((os.path.join(("./logs/" + folder), filename).replace('\\','/')).casefold()):
@@ -1888,17 +1939,11 @@ def match_index(sid0):
#print('nothing found')
return ('', '')
def choveraudio():
return ''
def stoptraining(mim):
if int(mim) == 1:
with open("stop.txt", "w+") as tostops:
tostops.writelines('stop')
cursor.execute("INSERT INTO stop_train (stop) VALUES (?)", (True,))
conn.commit()
#p.terminate()
#p.kill()
try:
@@ -1921,7 +1966,7 @@ def whethercrepeornah(radio):
return ({"visible": mango, "__type__": "update"})
#Change your Gradio Theme here. 👇 👇 👇 👇
#Change your Gradio Theme here. 👇 👇 👇 👇 Example: " theme='HaleyCH/HaleyCH_Theme' "
with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
gr.HTML("<h1> The Mangio-RVC-Fork 💻 </h1>")
gr.Markdown(
@@ -1977,10 +2022,10 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
input_audio1 = gr.Dropdown(
label=i18n("Auto detect audio path and select from the dropdown:"),
choices=sorted(audio_paths),
value=get_audios(),
value='',
interactive=True,
)
input_audio1.change(fn=choveraudio,inputs=[],outputs=[input_audio0])
input_audio1.change(fn=lambda:'',inputs=[],outputs=[input_audio0])
f0method0 = gr.Radio(
label=i18n(
"选择音高提取算法,输入歌声可用pm提速,harvest低音好但巨慢无比,crepe效果好但吃GPU"
@@ -2068,8 +2113,8 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
interactive=True,
)
formanting = gr.Checkbox(
value=False,
label="[EXPERIMENTAL, WAV ONLY] Formant shift inference audio",
value=bool(DoFormant),
label="[EXPERIMENTAL] Formant shift inference audio",
info="Used for male to female and vice-versa conversions",
interactive=True,
visible=True,
@@ -2079,11 +2124,14 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
value='',
choices=get_fshift_presets(),
label="browse presets for formanting",
visible=False,
visible=bool(DoFormant),
)
formant_refresh_button = gr.Button(
value='\U0001f504',
visible=bool(DoFormant),
variant='primary',
)
formant_refresh_button = gr.Button(value='\U0001f504', visible=False,variant='primary')
#formant_refresh_button = ToolButton( elem_id='1')
#create_refresh_button(formant_preset, lambda: {"choices": formant_preset}, "refresh_list_shiftpresets")
qfrency = gr.Slider(
value=Quefrency,
@@ -2092,9 +2140,10 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
minimum=0.0,
maximum=16.0,
step=0.1,
visible=False,
visible=bool(DoFormant),
interactive=True,
)
)
tmbre = gr.Slider(
value=Timbre,
info="Default value is 1.0",
@@ -2102,12 +2151,12 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
minimum=0.0,
maximum=16.0,
step=0.1,
visible=False,
visible=bool(DoFormant),
interactive=True,
)
formant_preset.change(fn=preset_apply, inputs=[formant_preset, qfrency, tmbre], outputs=[qfrency, tmbre])
frmntbut = gr.Button("Apply", variant="primary", visible=False)
frmntbut = gr.Button("Apply", variant="primary", visible=bool(DoFormant))
formanting.change(fn=formant_enabled,inputs=[formanting,qfrency,tmbre,frmntbut,formant_preset,formant_refresh_button],outputs=[formanting,qfrency,tmbre,frmntbut,formant_preset,formant_refresh_button])
frmntbut.click(fn=formant_apply,inputs=[qfrency, tmbre], outputs=[qfrency, tmbre])
formant_refresh_button.click(fn=update_fshift_presets,inputs=[formant_preset, qfrency, tmbre],outputs=[formant_preset, qfrency, tmbre])
@@ -2242,7 +2291,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
but1.click(
vc_multi,
[
sid0,
spk_item,
dir_input,
opt_input,
inputs,
@@ -2420,7 +2469,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
f0method8.change(fn=whethercrepeornah, inputs=[f0method8], outputs=[extraction_crepe_hop_length])
but2 = gr.Button(i18n("特征提取"), variant="primary")
info2 = gr.Textbox(label=i18n("输出信息"), value="", max_lines=8)
info2 = gr.Textbox(label=i18n("输出信息"), value="", max_lines=8, interactive=False)
but2.click(
extract_f0_feature,
[gpus6, np7, f0method8, if_f0_3, exp_dir1, version19, extraction_crepe_hop_length],
@@ -2455,8 +2504,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='Mangio-RVC-Web 💻') as app:
interactive=True,
)
if_save_latest13 = gr.Checkbox(
label="Whether to save only the latest .ckpt file to save hard disk space",
label="Whether to save only the latest .ckpt file to save hard drive space",
value=True,
interactive=True,
)

2
my_utils.py
View File

@@ -105,4 +105,4 @@ def load_audio(file, sr, DoFormant, Quefrency, Timbre):
converted = False
conn.close()
return np.frombuffer(out, np.float32).flatten()
return np.frombuffer(out, np.float32).flatten()

100
run.sh
View File

@@ -1,11 +1,5 @@
#!/bin/bash
# Define common paths for Homebrew
BREW_PATHS=(
"/usr/local/bin"
"/opt/homebrew/bin"
)
if [[ "$(uname)" == "Darwin" ]]; then
# macOS specific env:
export PYTORCH_ENABLE_MPS_FALLBACK=1
@@ -17,74 +11,6 @@ fi
requirements_file="requirements.txt"
# Function to add a path to PATH
add_to_path() {
echo "Homebrew found in $1, which is not in your PATH."
read -p "Do you want to add this path to your PATH? (y/n) " -n 1 -r
echo
if [[ $REPLY =~ ^[Yy]$ ]]; then
echo "Adding $1 to PATH..."
# Detect the shell and choose the right profile file
local shell_profile
if [[ $SHELL == *"/bash"* ]]; then
shell_profile="$HOME/.bashrc"
[[ ! -f "$shell_profile" ]] && shell_profile="$HOME/.bash_profile"
elif [[ $SHELL == *"/zsh"* ]]; then
shell_profile="$HOME/.zshrc"
else
echo "Unsupported shell. Please add the following line to your shell profile file manually:"
echo "export PATH=\"$PATH:$1\""
return
fi
# Add the export line to the shell profile file
echo "export PATH=\"$PATH:$1\"" >> "$shell_profile"
# Source the shell profile file
source "$shell_profile"
# Verify that the new PATH includes Homebrew
if ! command -v brew &> /dev/null; then
echo "Failed to add Homebrew to the PATH."
fi
fi
}
# Check if Homebrew is in PATH
if command -v brew &> /dev/null; then
echo "Homebrew is already in your PATH."
else
# If not, check common paths for Homebrew
echo "Homebrew not found in PATH. Checking common paths..."
for path in "${BREW_PATHS[@]}"; do
if [[ -x "$path/brew" ]]; then
add_to_path "$path"
break
fi
done
fi
# Check again if Homebrew is in PATH
if ! command -v brew &> /dev/null; then
echo "Homebrew still not found. Attempting to install..."
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
fi
# Verifying if Homebrew has been installed successfully
if command -v brew &> /dev/null; then
echo "Homebrew installed successfully."
else
echo "Homebrew installation failed."
exit 1
fi
# Installing ffmpeg with Homebrew
if [[ "$(uname)" == "Darwin" ]]; then
echo "Installing ffmpeg..."
brew install ffmpeg
fi
# Check if Python 3.8 is installed
if ! command -v python3.8 &> /dev/null; then
echo "Python 3.8 not found. Attempting to install..."
@@ -101,7 +27,7 @@ fi
# Check if required packages are installed and install them if not
if [ -f "${requirements_file}" ]; then
installed_packages=$(python3.8 -m pip list --format=freeze)
installed_packages=$(python3.8 -m pip freeze)
while IFS= read -r package; do
[[ "${package}" =~ ^#.* ]] && continue
package_name=$(echo "${package}" | sed 's/[<>=!].*//')
@@ -115,27 +41,5 @@ else
exit 1
fi
# Install onnxruntime package
echo "Installing onnxruntime..."
python3.8 -m pip install onnxruntime
download_if_not_exists() {
local filename=$1
local url=$2
if [ ! -f "$filename" ]; then
echo "$filename does not exist, downloading..."
curl -# -L -o "$filename" "$url"
echo "Download finished."
else
echo "$filename already exists."
fi
}
# Check and download hubert_base.pt
download_if_not_exists "hubert_base.pt" "https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/hubert_base.pt"
# Check and download rmvpe.pt
download_if_not_exists "rmvpe.pt" "https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/rmvpe.pt"
# Run the main script
python3.8 infer-web.py --pycmd python3.8
python3.8 infer-web.py --pycmd python3.8

0
stop.txt
View File

61
train_nsf_sim_cache_sid_load_pretrain.py
View File

@@ -33,6 +33,8 @@ from data_utils import (
DistributedBucketSampler,
)
import sqlite3
if hps.version == "v1":
from lib.infer_pack.models import (
SynthesizerTrnMs256NSFsid as RVC_Model_f0,
@@ -254,6 +256,7 @@ def run(rank, n_gpus, hps):
def train_and_evaluate(
rank, epoch, hps, nets, optims, schedulers, scaler, loaders, logger, writers, cache
):
net_g, net_d = nets
optim_g, optim_d = optims
train_loader, eval_loader = loaders
@@ -350,6 +353,10 @@ def train_and_evaluate(
# Run steps
epoch_recorder = EpochRecorder()
conn = sqlite3.connect('TEMP/db:cachedb?mode=memory&cache=shared', check_same_thread=False)
cursor = conn.cursor()
for batch_idx, info in data_iterator:
# Data
## Unpack
@@ -568,33 +575,33 @@ def train_and_evaluate(
),
)
)
with open("stop.txt", "r+") as tostop:
content = tostop.read()
if "stop" in content:
logger.info("Stop Button was pressed. The program is closed.")
if hasattr(net_g, "module"):
ckpt = net_g.module.state_dict()
else:
ckpt = net_g.state_dict()
logger.info(
"saving final ckpt:%s"
% (
savee(
ckpt,
hps.sample_rate,
hps.if_f0,
hps.name,
epoch,
hps.version,
hps,
)
cursor.execute("SELECT stop FROM stop_train LIMIT 1")
if bool(cursor.fetchone()) == True:
logger.info("Stop Button was pressed. The program is closed.")
if hasattr(net_g, "module"):
ckpt = net_g.module.state_dict()
else:
ckpt = net_g.state_dict()
logger.info(
"saving final ckpt:%s"
% (
savee(
ckpt,
hps.sample_rate,
hps.if_f0,
hps.name,
epoch,
hps.version,
hps,
)
)
tostop.truncate(0)
tostop.writelines("not")
os._exit(2333333)
)
sleep(1)
cursor.execute("DELETE FROM stop_train")
conn.commit()
conn.close()
os._exit(2333333)
if rank == 0:
logger.info("====> Epoch: {} {}".format(epoch, epoch_recorder.record()))
@@ -614,7 +621,11 @@ def train_and_evaluate(
)
)
sleep(1)
cursor.execute("DELETE FROM stop_train")
conn.commit()
conn.close()
os._exit(2333333)
if __name__ == "__main__":