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

Merge pull request #65 from Mangio621/fix-cuda-req

Browse Source 
Undo torch requirement change for compatibility
This commit is contained in:
kalomaze
2023-07-28 18:47:13 -05:00
committed by GitHub
parent e2b3aa0f3b 93706e95f5
commit ed9c9cda8c
5 changed files with 596 additions and 353 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
extract_f0_print.py
View File

@@ -22,6 +22,7 @@ DoFormant = False
Quefrency = 0.0
Timbre = 0.0
def printt(strr):
print(strr)
f.write("%s\n" % strr)

550
infer-web.py
View File

File diff suppressed because it is too large Load Diff

84
my_utils.py
View File

@@ -11,22 +11,25 @@ import random
import csv
platform_stft_mapping = {
'linux': 'stftpitchshift',
'darwin': 'stftpitchshift',
'win32': 'stftpitchshift.exe',
"linux": "stftpitchshift",
"darwin": "stftpitchshift",
"win32": "stftpitchshift.exe",
}
stft = platform_stft_mapping.get(sys.platform)
# praatEXE = join('.',os.path.abspath(os.getcwd()) + r"\Praat.exe")
def CSVutil(file, rw, type, *args):
if type == 'formanting':
if rw == 'r':
if type == "formanting":
if rw == "r":
with open(file) as fileCSVread:
csv_reader = list(csv.reader(fileCSVread))
return (
csv_reader[0][0], csv_reader[0][1], csv_reader[0][2]
) if csv_reader is not None else (lambda: exec('raise ValueError("No data")'))()
(csv_reader[0][0], csv_reader[0][1], csv_reader[0][2])
if csv_reader is not None
else (lambda: exec('raise ValueError("No data")'))()
)
else:
if args:
doformnt = args[0]
@@ -34,18 +37,19 @@ def CSVutil(file, rw, type, *args):
doformnt = False
qfr = args[1] if len(args) > 1 else 1.0
tmb = args[2] if len(args) > 2 else 1.0
with open(file, rw, newline='') as fileCSVwrite:
csv_writer = csv.writer(fileCSVwrite, delimiter=',')
with open(file, rw, newline="") as fileCSVwrite:
csv_writer = csv.writer(fileCSVwrite, delimiter=",")
csv_writer.writerow([doformnt, qfr, tmb])
elif type == 'stop':
elif type == "stop":
stop = args[0] if args else False
with open(file, rw, newline='') as fileCSVwrite:
csv_writer = csv.writer(fileCSVwrite, delimiter=',')
with open(file, rw, newline="") as fileCSVwrite:
csv_writer = csv.writer(fileCSVwrite, delimiter=",")
csv_writer.writerow([stop])
def load_audio(file, sr, DoFormant, Quefrency, Timbre):
converted = False
DoFormant, Quefrency, Timbre = CSVutil('csvdb/formanting.csv', 'r', 'formanting')
DoFormant, Quefrency, Timbre = CSVutil("csvdb/formanting.csv", "r", "formanting")
try:
# https://github.com/openai/whisper/blob/main/whisper/audio.py#L26
# This launches a subprocess to decode audio while down-mixing and resampling as necessary.
@@ -57,13 +61,16 @@ def load_audio(file, sr, DoFormant, Quefrency, Timbre):
# print(f"dofor={bool(DoFormant)} timbr={Timbre} quef={Quefrency}\n")
if (lambda DoFormant: True if DoFormant.lower() == 'true' else (False if DoFormant.lower() == 'false' else DoFormant))(DoFormant):
if (
lambda DoFormant: True
if DoFormant.lower() == "true"
else (False if DoFormant.lower() == "false" else DoFormant)
)(DoFormant):
numerator = round(random.uniform(1, 4), 4)
# os.system(f"stftpitchshift -i {file} -q {Quefrency} -t {Timbre} -o {file_formanted}")
# print('stftpitchshift -i "%s" -p 1.0 --rms -w 128 -v 8 -q %s -t %s -o "%s"' % (file, Quefrency, Timbre, file_formanted))
if not file.endswith(".wav"):
if not os.path.isfile(f"{file_formanted}.wav"):
converted = True
# print(f"\nfile = {file}\n")
@@ -72,47 +79,55 @@ def load_audio(file, sr, DoFormant, Quefrency, Timbre):
ffmpeg.input(file_formanted, threads=0)
.output(f"{file_formanted}.wav")
.run(
cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True
cmd=["ffmpeg", "-nostdin"],
capture_stdout=True,
capture_stderr=True,
)
)
else:
pass
file_formanted = f"{file_formanted}.wav" if not file_formanted.endswith(".wav") else file_formanted
file_formanted = (
f"{file_formanted}.wav"
if not file_formanted.endswith(".wav")
else file_formanted
)
print(f" · Formanting {file_formanted}...\n")
os.system(
'%s -i "%s" -q "%s" -t "%s" -o "%sFORMANTED_%s.wav"'
% (stft, file_formanted, Quefrency, Timbre, file_formanted, str(numerator))
% (
stft,
file_formanted,
Quefrency,
Timbre,
file_formanted,
str(numerator),
)
)
print(f" · Formanted {file_formanted}!\n")
# filepraat = (os.path.abspath(os.getcwd()) + '\\' + file).replace('/','\\')
# file_formantedpraat = ('"' + os.path.abspath(os.getcwd()) + '/' + 'formanted'.join(file_formanted) + '"').replace('/','\\')
# print("%sFORMANTED_%s.wav" % (file_formanted, str(numerator)))
out, _ = (
ffmpeg.input("%sFORMANTED_%s.wav" % (file_formanted, str(numerator)), threads=0)
ffmpeg.input(
"%sFORMANTED_%s.wav" % (file_formanted, str(numerator)), threads=0
)
.output("-", format="f32le", acodec="pcm_f32le", ac=1, ar=sr)
.run(
cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True
)
)
try: os.remove("%sFORMANTED_%s.wav" % (file_formanted, str(numerator)))
except Exception: pass; print("couldn't remove formanted type of file")
try:
os.remove("%sFORMANTED_%s.wav" % (file_formanted, str(numerator)))
except Exception:
pass
print("couldn't remove formanted type of file")
else:
out, _ = (
@@ -126,8 +141,11 @@ def load_audio(file, sr, DoFormant, Quefrency, Timbre):
raise RuntimeError(f"Failed to load audio: {e}")
if converted:
try: os.remove(file_formanted)
except Exception: pass; print("couldn't remove converted type of file")
try:
os.remove(file_formanted)
except Exception:
pass
print("couldn't remove converted type of file")
converted = False
return np.frombuffer(out, np.float32).flatten()

2
requirements.txt
View File

@@ -146,7 +146,7 @@ tensorboard-plugin-wit==1.8.1
tensorboardX==2.6.1
threadpoolctl==3.1.0
toolz==0.12.0
torch==2.0.1
torch @ https://download.pytorch.org/whl/cu118/torch-2.0.0%2Bcu118-cp39-cp39-win_amd64.whl
torchaudio==2.0.1
torchcrepe==0.0.19
torchgen==0.0.1

28
train_nsf_sim_cache_sid_load_pretrain.py
View File

@@ -256,7 +256,6 @@ 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
@@ -574,10 +573,18 @@ def train_and_evaluate(
)
try:
with open('csvdb/stop.csv') as CSVStop:
with open("csvdb/stop.csv") as CSVStop:
csv_reader = list(csv.reader(CSVStop))
stopbtn = csv_reader[0][0] if csv_reader is not None else (lambda: exec('raise ValueError("No data")'))()
stopbtn = (lambda stopbtn: True if stopbtn.lower() == 'true' else (False if stopbtn.lower() == 'false' else stopbtn))(stopbtn)
stopbtn = (
csv_reader[0][0]
if csv_reader is not None
else (lambda: exec('raise ValueError("No data")'))()
)
stopbtn = (
lambda stopbtn: True
if stopbtn.lower() == "true"
else (False if stopbtn.lower() == "false" else stopbtn)
)(stopbtn)
except (ValueError, TypeError, IndexError):
stopbtn = False
@@ -602,9 +609,9 @@ def train_and_evaluate(
)
)
sleep(1)
with open('csvdb/stop.csv', 'w+', newline='') as STOPCSVwrite:
csv_writer = csv.writer(STOPCSVwrite, delimiter=',')
csv_writer.writerow(['False'])
with open("csvdb/stop.csv", "w+", newline="") as STOPCSVwrite:
csv_writer = csv.writer(STOPCSVwrite, delimiter=",")
csv_writer.writerow(["False"])
os._exit(2333333)
if rank == 0:
@@ -625,13 +632,12 @@ def train_and_evaluate(
)
)
sleep(1)
with open('csvdb/stop.csv', 'w+', newline='') as STOPCSVwrite:
csv_writer = csv.writer(STOPCSVwrite, delimiter=',')
csv_writer.writerow(['False'])
with open("csvdb/stop.csv", "w+", newline="") as STOPCSVwrite:
csv_writer = csv.writer(STOPCSVwrite, delimiter=",")
csv_writer.writerow(["False"])
os._exit(2333333)
if __name__ == "__main__":
torch.multiprocessing.set_start_method("spawn")
main()