mirror of
https://github.com/Anjok07/ultimatevocalremovergui.git
synced 2024-09-24 03:28:36 +02:00
Add files via upload
This commit is contained in:
Anjok07
GitHub
parent
063f015d5d
commit
b011d37a58
995
inference_MDX.py
995
inference_MDX.py
File diff suppressed because it is too large
Load Diff
@ -1,37 +1,37 @@
|
||||
import os
|
||||
from pathlib import Path
|
||||
import os.path
|
||||
from datetime import datetime
|
||||
import pydub
|
||||
import shutil
|
||||
from random import randrange
|
||||
#MDX-Net
|
||||
#----------------------------------------
|
||||
import soundfile as sf
|
||||
import torch
|
||||
import numpy as np
|
||||
|
||||
from demucs.pretrained import get_model as _gm
|
||||
from demucs.hdemucs import HDemucs
|
||||
from demucs.apply import BagOfModels, apply_model
|
||||
from demucs.audio import AudioFile
|
||||
|
||||
import time
|
||||
import os
|
||||
from tqdm import tqdm
|
||||
import warnings
|
||||
import sys
|
||||
import librosa
|
||||
import psutil
|
||||
#----------------------------------------
|
||||
from demucs.hdemucs import HDemucs
|
||||
from demucs.model_v2 import Demucs
|
||||
from demucs.pretrained import get_model as _gm
|
||||
from demucs.tasnet_v2 import ConvTasNet
|
||||
from demucs.utils import apply_model_v1
|
||||
from demucs.utils import apply_model_v2
|
||||
from diffq import DiffQuantizer
|
||||
from lib_v5 import spec_utils
|
||||
from lib_v5.model_param_init import ModelParameters
|
||||
import torch
|
||||
|
||||
# Command line text parsing and widget manipulation
|
||||
import tkinter as tk
|
||||
import traceback # Error Message Recent Calls
|
||||
from pathlib import Path
|
||||
from random import randrange
|
||||
from tqdm import tqdm
|
||||
import gzip
|
||||
import io
|
||||
import librosa
|
||||
import numpy as np
|
||||
import os
|
||||
import os
|
||||
import os.path
|
||||
import psutil
|
||||
import pydub
|
||||
import shutil
|
||||
import soundfile as sf
|
||||
import sys
|
||||
import time
|
||||
import time # Timer
|
||||
import tkinter as tk
|
||||
import torch
|
||||
import torch.hub
|
||||
import traceback # Error Message Recent Calls
|
||||
import warnings
|
||||
import zlib
|
||||
|
||||
class Predictor():
|
||||
def __init__(self):
|
||||
@ -46,40 +46,62 @@ class Predictor():
|
||||
if data['gpu'] == -1:
|
||||
device = torch.device('cpu')
|
||||
|
||||
self.demucs = HDemucs(sources=["drums", "bass", "other", "vocals"])
|
||||
widget_text.write(base_text + 'Loading Demucs model... ')
|
||||
update_progress(**progress_kwargs,
|
||||
step=0.05)
|
||||
path_d = Path('models/Demucs_Models')
|
||||
print('What Demucs model was chosen? ', data['DemucsModel'])
|
||||
self.demucs = _gm(name=data['DemucsModel'], repo=path_d)
|
||||
widget_text.write('Done!\n')
|
||||
if 'UVR' in data['DemucsModel']:
|
||||
widget_text.write(base_text + "2 stem model selected.\n")
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
widget_text.write(base_text + f"Selected model is a bag of {len(self.demucs.models)} models.\n")
|
||||
|
||||
if data['segment'] == 'None':
|
||||
segment = None
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
if segment is not None:
|
||||
for sub in self.demucs.models:
|
||||
sub.segment = segment
|
||||
if demucs_model_version == 'v1':
|
||||
load_from = "models/Demucs_Models/"f"{demucs_model_set_name}"
|
||||
if str(load_from).endswith(".gz"):
|
||||
load_from = gzip.open(load_from, "rb")
|
||||
klass, args, kwargs, state = torch.load(load_from)
|
||||
self.demucs = klass(*args, **kwargs)
|
||||
widget_text.write(base_text + 'Loading Demucs v1 model... ')
|
||||
update_progress(**progress_kwargs,
|
||||
step=0.05)
|
||||
self.demucs.to(device)
|
||||
self.demucs.load_state_dict(state)
|
||||
widget_text.write('Done!\n')
|
||||
if not data['segment'] == 'None':
|
||||
widget_text.write(base_text + 'Segments is only available in Demucs v3. Please use \"Chunks\" instead.\n')
|
||||
else:
|
||||
if segment is not None:
|
||||
sub.segment = segment
|
||||
else:
|
||||
try:
|
||||
segment = int(data['segment'])
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
if segment is not None:
|
||||
for sub in self.demucs.models:
|
||||
sub.segment = segment
|
||||
else:
|
||||
if segment is not None:
|
||||
sub.segment = segment
|
||||
widget_text.write(base_text + "Segments set to "f"{segment}.\n")
|
||||
except:
|
||||
pass
|
||||
|
||||
if demucs_model_version == 'v2':
|
||||
if '48' in demucs_model_set_name:
|
||||
channels=48
|
||||
elif 'unittest' in demucs_model_set_name:
|
||||
channels=4
|
||||
else:
|
||||
channels=64
|
||||
|
||||
if 'tasnet' in demucs_model_set_name:
|
||||
self.demucs = ConvTasNet(sources=["drums", "bass", "other", "vocals"], X=10)
|
||||
else:
|
||||
self.demucs = Demucs(sources=["drums", "bass", "other", "vocals"], channels=channels)
|
||||
widget_text.write(base_text + 'Loading Demucs v2 model... ')
|
||||
update_progress(**progress_kwargs,
|
||||
step=0.05)
|
||||
self.demucs.to(device)
|
||||
self.demucs.load_state_dict(torch.load("models/Demucs_Models/"f"{demucs_model_set_name}"))
|
||||
widget_text.write('Done!\n')
|
||||
if not data['segment'] == 'None':
|
||||
widget_text.write(base_text + 'Segments is only available in Demucs v3. Please use \"Chunks\" instead.\n')
|
||||
else:
|
||||
pass
|
||||
self.demucs.eval()
|
||||
|
||||
if demucs_model_version == 'v3':
|
||||
self.demucs = HDemucs(sources=["drums", "bass", "other", "vocals"])
|
||||
widget_text.write(base_text + 'Loading Demucs model... ')
|
||||
update_progress(**progress_kwargs,
|
||||
step=0.05)
|
||||
path_d = Path('models/Demucs_Models/v3_repo')
|
||||
#print('What Demucs model was chosen? ', demucs_model_set_name)
|
||||
self.demucs = _gm(name=demucs_model_set_name, repo=path_d)
|
||||
widget_text.write('Done!\n')
|
||||
if 'UVR' in data['DemucsModel']:
|
||||
widget_text.write(base_text + "2 stem model selected.\n")
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
widget_text.write(base_text + f"Selected model is a bag of {len(self.demucs.models)} models.\n")
|
||||
|
||||
if data['segment'] == 'None':
|
||||
segment = None
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
if segment is not None:
|
||||
@ -88,9 +110,29 @@ class Predictor():
|
||||
else:
|
||||
if segment is not None:
|
||||
sub.segment = segment
|
||||
|
||||
self.demucs.to(device)
|
||||
self.demucs.eval()
|
||||
else:
|
||||
try:
|
||||
segment = int(data['segment'])
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
if segment is not None:
|
||||
for sub in self.demucs.models:
|
||||
sub.segment = segment
|
||||
else:
|
||||
if segment is not None:
|
||||
sub.segment = segment
|
||||
widget_text.write(base_text + "Segments set to "f"{segment}.\n")
|
||||
except:
|
||||
segment = None
|
||||
if isinstance(self.demucs, BagOfModels):
|
||||
if segment is not None:
|
||||
for sub in self.demucs.models:
|
||||
sub.segment = segment
|
||||
else:
|
||||
if segment is not None:
|
||||
sub.segment = segment
|
||||
|
||||
self.demucs.to(device)
|
||||
self.demucs.eval()
|
||||
|
||||
update_progress(**progress_kwargs,
|
||||
step=0.1)
|
||||
@ -646,7 +688,12 @@ class Predictor():
|
||||
if end == samples:
|
||||
break
|
||||
|
||||
sources = self.demix_demucs(segmented_mix, margin_size=margin)
|
||||
if demucs_model_version == 'v1':
|
||||
sources = self.demix_demucs_v1(segmented_mix, margin_size=margin)
|
||||
if demucs_model_version == 'v2':
|
||||
sources = self.demix_demucs_v2(segmented_mix, margin_size=margin)
|
||||
if demucs_model_version == 'v3':
|
||||
sources = self.demix_demucs(segmented_mix, margin_size=margin)
|
||||
|
||||
return sources
|
||||
|
||||
@ -683,31 +730,94 @@ class Predictor():
|
||||
sources = np.concatenate(sources, axis=-1)
|
||||
widget_text.write('Done!\n')
|
||||
return sources
|
||||
|
||||
def demix_demucs_v1(self, mix, margin_size):
|
||||
processed = {}
|
||||
demucsitera = len(mix)
|
||||
demucsitera_calc = demucsitera * 2
|
||||
gui_progress_bar_demucs = 0
|
||||
widget_text.write(base_text + "Running Demucs v1 Inference...\n")
|
||||
widget_text.write(base_text + "Processing "f"{len(mix)} slices... ")
|
||||
print(' Running Demucs Inference...')
|
||||
for nmix in mix:
|
||||
gui_progress_bar_demucs += 1
|
||||
update_progress(**progress_kwargs,
|
||||
step=(0.35 + (1.05/demucsitera_calc * gui_progress_bar_demucs)))
|
||||
cmix = mix[nmix]
|
||||
cmix = torch.tensor(cmix, dtype=torch.float32)
|
||||
ref = cmix.mean(0)
|
||||
cmix = (cmix - ref.mean()) / ref.std()
|
||||
with torch.no_grad():
|
||||
sources = apply_model_v1(self.demucs, cmix.to(device), split=split_mode, shifts=shift_set)
|
||||
sources = (sources * ref.std() + ref.mean()).cpu().numpy()
|
||||
sources[[0,1]] = sources[[1,0]]
|
||||
|
||||
start = 0 if nmix == 0 else margin_size
|
||||
end = None if nmix == list(mix.keys())[::-1][0] else -margin_size
|
||||
if margin_size == 0:
|
||||
end = None
|
||||
processed[nmix] = sources[:,:,start:end].copy()
|
||||
|
||||
sources = list(processed.values())
|
||||
sources = np.concatenate(sources, axis=-1)
|
||||
widget_text.write('Done!\n')
|
||||
return sources
|
||||
|
||||
def demix_demucs_v2(self, mix, margin_size):
|
||||
processed = {}
|
||||
demucsitera = len(mix)
|
||||
demucsitera_calc = demucsitera * 2
|
||||
gui_progress_bar_demucs = 0
|
||||
widget_text.write(base_text + "Running Demucs v2 Inference...\n")
|
||||
widget_text.write(base_text + "Processing "f"{len(mix)} slices... ")
|
||||
print(' Running Demucs Inference...')
|
||||
for nmix in mix:
|
||||
gui_progress_bar_demucs += 1
|
||||
update_progress(**progress_kwargs,
|
||||
step=(0.35 + (1.05/demucsitera_calc * gui_progress_bar_demucs)))
|
||||
cmix = mix[nmix]
|
||||
cmix = torch.tensor(cmix, dtype=torch.float32)
|
||||
ref = cmix.mean(0)
|
||||
cmix = (cmix - ref.mean()) / ref.std()
|
||||
shift_set = 0
|
||||
with torch.no_grad():
|
||||
sources = apply_model_v2(self.demucs, cmix.to(device), split=split_mode, overlap=overlap_set, shifts=shift_set)
|
||||
sources = (sources * ref.std() + ref.mean()).cpu().numpy()
|
||||
sources[[0,1]] = sources[[1,0]]
|
||||
|
||||
start = 0 if nmix == 0 else margin_size
|
||||
end = None if nmix == list(mix.keys())[::-1][0] else -margin_size
|
||||
if margin_size == 0:
|
||||
end = None
|
||||
processed[nmix] = sources[:,:,start:end].copy()
|
||||
|
||||
sources = list(processed.values())
|
||||
sources = np.concatenate(sources, axis=-1)
|
||||
widget_text.write('Done!\n')
|
||||
return sources
|
||||
|
||||
data = {
|
||||
# Paths
|
||||
'input_paths': None,
|
||||
'export_path': None,
|
||||
'saveFormat': 'Wav',
|
||||
# Processing Options
|
||||
'demucsmodel': True,
|
||||
'gpu': -1,
|
||||
'audfile': True,
|
||||
'chunks_d': 'Full',
|
||||
'settest': False,
|
||||
'voc_only_b': False,
|
||||
'inst_only_b': False,
|
||||
'overlap_b': 0.25,
|
||||
'shifts_b': 2,
|
||||
'segment': 'None',
|
||||
'margin': 44100,
|
||||
'split_mode': False,
|
||||
'normalize': False,
|
||||
'compensate': 1.03597672895,
|
||||
'demucs_stems': 'All Stems',
|
||||
'DemucsModel': 'mdx_extra',
|
||||
'audfile': True,
|
||||
'wavtype': 'PCM_16',
|
||||
'demucsmodel': True,
|
||||
'export_path': None,
|
||||
'gpu': -1,
|
||||
'input_paths': None,
|
||||
'inst_only_b': False,
|
||||
'margin': 44100,
|
||||
'mp3bit': '320k',
|
||||
'normalize': False,
|
||||
'overlap_b': 0.25,
|
||||
'saveFormat': 'Wav',
|
||||
'segment': 'None',
|
||||
'settest': False,
|
||||
'shifts_b': 2,
|
||||
'split_mode': False,
|
||||
'voc_only_b': False,
|
||||
'wavtype': 'PCM_16',
|
||||
}
|
||||
default_chunks = data['chunks_d']
|
||||
|
||||
@ -756,6 +866,8 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
global shift_set
|
||||
global source_val
|
||||
global split_mode
|
||||
global demucs_model_set_name
|
||||
global demucs_model_version
|
||||
|
||||
global wav_type_set
|
||||
global flac_type_set
|
||||
@ -817,6 +929,69 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
progress_var.set(0)
|
||||
text_widget.clear()
|
||||
button_widget.configure(state=tk.DISABLED) # Disable Button
|
||||
|
||||
|
||||
if data['DemucsModel'] == "Tasnet v1":
|
||||
demucs_model_set_name = 'tasnet.th'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Tasnet_extra v1":
|
||||
demucs_model_set_name = 'tasnet_extra.th'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Demucs v1":
|
||||
demucs_model_set_name = 'demucs.th'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Demucs v1.gz":
|
||||
demucs_model_set_name = 'demucs.th.gz'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Demucs_extra v1":
|
||||
demucs_model_set_name = 'demucs_extra.th'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Demucs_extra v1.gz":
|
||||
demucs_model_set_name = 'demucs_extra.th.gz'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Light v1":
|
||||
demucs_model_set_name = 'light.th'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Light v1.gz":
|
||||
demucs_model_set_name = 'light.th.gz'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Light_extra v1":
|
||||
demucs_model_set_name = 'light_extra.th'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Light_extra v1.gz":
|
||||
demucs_model_set_name = 'light_extra.th.gz'
|
||||
demucs_model_version = 'v1'
|
||||
elif data['DemucsModel'] == "Tasnet v2":
|
||||
demucs_model_set_name = 'tasnet-beb46fac.th'
|
||||
demucs_model_version = 'v2'
|
||||
elif data['DemucsModel'] == "Tasnet_extra v2":
|
||||
demucs_model_set_name = 'tasnet_extra-df3777b2.th'
|
||||
demucs_model_version = 'v2'
|
||||
elif data['DemucsModel'] == "Demucs48_hq v2":
|
||||
demucs_model_set_name = 'demucs48_hq-28a1282c.th'
|
||||
demucs_model_version = 'v2'
|
||||
elif data['DemucsModel'] == "Demucs v2":
|
||||
demucs_model_set_name = 'demucs-e07c671f.th'
|
||||
demucs_model_version = 'v2'
|
||||
elif data['DemucsModel'] == "Demucs_extra v2":
|
||||
demucs_model_set_name = 'demucs_extra-3646af93.th'
|
||||
demucs_model_version = 'v2'
|
||||
elif data['DemucsModel'] == "Demucs_unittest v2":
|
||||
demucs_model_set_name = 'demucs_unittest-09ebc15f.th'
|
||||
demucs_model_version = 'v2'
|
||||
elif '.ckpt' in data['DemucsModel'] and 'v2' in data['DemucsModel']:
|
||||
demucs_model_set_name = data['DemucsModel']
|
||||
demucs_model_version = 'v2'
|
||||
elif '.ckpt' in data['DemucsModel'] and 'v1' in data['DemucsModel']:
|
||||
demucs_model_set_name = data['DemucsModel']
|
||||
demucs_model_version = 'v1'
|
||||
elif '.gz' in data['DemucsModel']:
|
||||
demucs_model_set_name = data['DemucsModel']
|
||||
demucs_model_version = 'v1'
|
||||
else:
|
||||
demucs_model_set_name = data['DemucsModel']
|
||||
demucs_model_version = 'v3'
|
||||
|
||||
|
||||
try: #Load File(s)
|
||||
for file_num, music_file in tqdm(enumerate(data['input_paths'], start=1)):
|
||||
@ -880,7 +1055,10 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
channel_set = int(data['channel'])
|
||||
margin_set = int(data['margin'])
|
||||
shift_set = int(data['shifts_b'])
|
||||
|
||||
split_mode = data['split_mode']
|
||||
|
||||
#print('Split? ', split_mode)
|
||||
|
||||
def determinemusicfileFolderName():
|
||||
"""
|
||||
|
||||
380
inference_v5.py
380
inference_v5.py
@ -1,34 +1,31 @@
|
||||
import os
|
||||
import importlib
|
||||
import pydub
|
||||
import shutil
|
||||
import hashlib
|
||||
|
||||
import cv2
|
||||
import librosa
|
||||
import math
|
||||
import numpy as np
|
||||
import soundfile as sf
|
||||
from tqdm import tqdm
|
||||
|
||||
from demucs.pretrained import get_model as _gm
|
||||
from demucs.hdemucs import HDemucs
|
||||
from collections import defaultdict
|
||||
from datetime import datetime
|
||||
from demucs.apply import BagOfModels, apply_model
|
||||
from pathlib import Path
|
||||
from models import stft, istft
|
||||
|
||||
from demucs.hdemucs import HDemucs
|
||||
from demucs.pretrained import get_model as _gm
|
||||
from lib_v5 import dataset
|
||||
from lib_v5 import spec_utils
|
||||
from lib_v5.model_param_init import ModelParameters
|
||||
import torch
|
||||
from datetime import datetime
|
||||
|
||||
# Command line text parsing and widget manipulation
|
||||
from collections import defaultdict
|
||||
import tkinter as tk
|
||||
import traceback # Error Message Recent Calls
|
||||
import time # Timer
|
||||
from models import stft, istft
|
||||
from pathlib import Path
|
||||
from random import randrange
|
||||
from tqdm import tqdm
|
||||
from tkinter import filedialog
|
||||
import lib_v5.filelist
|
||||
import cv2
|
||||
import hashlib
|
||||
import importlib
|
||||
import librosa
|
||||
import math
|
||||
import numpy as np
|
||||
import os
|
||||
import pydub
|
||||
import shutil
|
||||
import soundfile as sf
|
||||
import time # Timer
|
||||
import tkinter as tk
|
||||
import torch
|
||||
import traceback # Error Message Recent Calls
|
||||
|
||||
class VocalRemover(object):
|
||||
|
||||
@ -40,35 +37,31 @@ class VocalRemover(object):
|
||||
# self.offset = model.offset
|
||||
|
||||
data = {
|
||||
# Paths
|
||||
'input_paths': None,
|
||||
'export_path': None,
|
||||
'saveFormat': 'wav',
|
||||
# Processing Options
|
||||
'gpu': -1,
|
||||
'postprocess': True,
|
||||
'tta': True,
|
||||
'output_image': True,
|
||||
'voc_only': False,
|
||||
'inst_only': False,
|
||||
# Models
|
||||
'instrumentalModel': None,
|
||||
'useModel': None,
|
||||
# Constants
|
||||
'window_size': 512,
|
||||
'agg': 10,
|
||||
'high_end_process': 'mirroring',
|
||||
'ModelParams': 'Auto',
|
||||
'demucsmodel_sel_VR': 'UVR_Demucs_Model_1',
|
||||
'overlap': 0.5,
|
||||
'shifts': 0,
|
||||
'segment': 'None',
|
||||
'split_mode': False,
|
||||
'normalize': False,
|
||||
'demucsmodelVR': True,
|
||||
'wavtype': 'PCM_16',
|
||||
'export_path': None,
|
||||
'gpu': -1,
|
||||
'high_end_process': 'mirroring',
|
||||
'input_paths': None,
|
||||
'inst_only': False,
|
||||
'instrumentalModel': None,
|
||||
'ModelParams': 'Auto',
|
||||
'mp3bit': '320k',
|
||||
'normalize': False,
|
||||
'output_image': True,
|
||||
'overlap': 0.5,
|
||||
'postprocess': True,
|
||||
'saveFormat': 'wav',
|
||||
'segment': 'None',
|
||||
'settest': False,
|
||||
'shifts': 0,
|
||||
'split_mode': False,
|
||||
'tta': True,
|
||||
'useModel': None,
|
||||
'voc_only': False,
|
||||
'wavtype': 'PCM_16',
|
||||
'window_size': 512,
|
||||
}
|
||||
|
||||
default_window_size = data['window_size']
|
||||
@ -144,7 +137,7 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
|
||||
nn_arch_sizes = [
|
||||
31191, # default
|
||||
33966, 123821, 123812, 537238 # custom
|
||||
33966, 123821, 123812, 129605, 537238 # custom
|
||||
]
|
||||
|
||||
nn_architecture = list('{}KB'.format(s) for s in nn_arch_sizes)
|
||||
@ -492,7 +485,7 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
return
|
||||
|
||||
#Load Model
|
||||
text_widget.write(base_text + 'Loading models...')
|
||||
text_widget.write(base_text + 'Loading model...')
|
||||
|
||||
model_size = math.ceil(os.stat(data['instrumentalModel']).st_size / 1024)
|
||||
nn_architecture = '{}KB'.format(min(nn_arch_sizes, key=lambda x:abs(x-model_size)))
|
||||
@ -504,212 +497,77 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
ModelName=(data['instrumentalModel'])
|
||||
|
||||
#Package Models
|
||||
|
||||
model_hash = hashlib.md5(open(ModelName,'rb').read()).hexdigest()
|
||||
print(model_hash)
|
||||
|
||||
#v5 Models
|
||||
|
||||
if model_hash == '47939caf0cfe52a0e81442b85b971dfd':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100')
|
||||
if model_hash == '4e4ecb9764c50a8c414fee6e10395bbe':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_v2.json')
|
||||
param_name_auto=str('4band_v2')
|
||||
if model_hash == 'e60a1e84803ce4efc0a6551206cc4b71':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100')
|
||||
if model_hash == 'a82f14e75892e55e994376edbf0c8435':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100')
|
||||
if model_hash == '6dd9eaa6f0420af9f1d403aaafa4cc06':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_v2_sn.json')
|
||||
param_name_auto=str('4band_v2_sn')
|
||||
if model_hash == '5c7bbca45a187e81abbbd351606164e5':
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100_msb2.json')
|
||||
param_name_auto=str('3band_44100_msb2')
|
||||
if model_hash == 'd6b2cb685a058a091e5e7098192d3233':
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100_msb2.json')
|
||||
param_name_auto=str('3band_44100_msb2')
|
||||
if model_hash == 'c1b9f38170a7c90e96f027992eb7c62b':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100')
|
||||
if model_hash == 'c3448ec923fa0edf3d03a19e633faa53':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100')
|
||||
if model_hash == '68aa2c8093d0080704b200d140f59e54':
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100.json')
|
||||
param_name_auto=str('3band_44100.json')
|
||||
if model_hash == 'fdc83be5b798e4bd29fe00fe6600e147':
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100_mid.json')
|
||||
param_name_auto=str('3band_44100_mid.json')
|
||||
if model_hash == '2ce34bc92fd57f55db16b7a4def3d745':
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100_mid.json')
|
||||
param_name_auto=str('3band_44100_mid.json')
|
||||
if model_hash == '52fdca89576f06cf4340b74a4730ee5f':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100.json')
|
||||
if model_hash == '41191165b05d38fc77f072fa9e8e8a30':
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100.json')
|
||||
if model_hash == '89e83b511ad474592689e562d5b1f80e':
|
||||
model_params_auto=str('lib_v5/modelparams/2band_32000.json')
|
||||
param_name_auto=str('2band_32000.json')
|
||||
if model_hash == '0b954da81d453b716b114d6d7c95177f':
|
||||
model_params_auto=str('lib_v5/modelparams/2band_32000.json')
|
||||
param_name_auto=str('2band_32000.json')
|
||||
|
||||
#v4 Models
|
||||
|
||||
if model_hash == '6a00461c51c2920fd68937d4609ed6c8':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr16000_hl512.json')
|
||||
param_name_auto=str('1band_sr16000_hl512')
|
||||
if model_hash == '0ab504864d20f1bd378fe9c81ef37140':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr32000_hl512.json')
|
||||
param_name_auto=str('1band_sr32000_hl512')
|
||||
if model_hash == '7dd21065bf91c10f7fccb57d7d83b07f':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr32000_hl512.json')
|
||||
param_name_auto=str('1band_sr32000_hl512')
|
||||
if model_hash == '80ab74d65e515caa3622728d2de07d23':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr32000_hl512.json')
|
||||
param_name_auto=str('1band_sr32000_hl512')
|
||||
if model_hash == 'edc115e7fc523245062200c00caa847f':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr33075_hl384.json')
|
||||
param_name_auto=str('1band_sr33075_hl384')
|
||||
if model_hash == '28063e9f6ab5b341c5f6d3c67f2045b7':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr33075_hl384.json')
|
||||
param_name_auto=str('1band_sr33075_hl384')
|
||||
if model_hash == 'b58090534c52cbc3e9b5104bad666ef2':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr44100_hl512.json')
|
||||
param_name_auto=str('1band_sr44100_hl512')
|
||||
if model_hash == '0cdab9947f1b0928705f518f3c78ea8f':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr44100_hl512.json')
|
||||
param_name_auto=str('1band_sr44100_hl512')
|
||||
if model_hash == 'ae702fed0238afb5346db8356fe25f13':
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr44100_hl1024.json')
|
||||
param_name_auto=str('1band_sr44100_hl1024')
|
||||
|
||||
#User Models
|
||||
|
||||
#1 Band
|
||||
if '1band_sr16000_hl512' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr16000_hl512.json')
|
||||
param_name_auto=str('1band_sr16000_hl512')
|
||||
if '1band_sr32000_hl512' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr32000_hl512.json')
|
||||
param_name_auto=str('1band_sr32000_hl512')
|
||||
if '1band_sr33075_hl384' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr33075_hl384.json')
|
||||
param_name_auto=str('1band_sr33075_hl384')
|
||||
if '1band_sr44100_hl256' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr44100_hl256.json')
|
||||
param_name_auto=str('1band_sr44100_hl256')
|
||||
if '1band_sr44100_hl512' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr44100_hl512.json')
|
||||
param_name_auto=str('1band_sr44100_hl512')
|
||||
if '1band_sr44100_hl1024' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/1band_sr44100_hl1024.json')
|
||||
param_name_auto=str('1band_sr44100_hl1024')
|
||||
|
||||
#2 Band
|
||||
if '2band_44100_lofi' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/2band_44100_lofi.json')
|
||||
param_name_auto=str('2band_44100_lofi')
|
||||
if '2band_32000' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/2band_32000.json')
|
||||
param_name_auto=str('2band_32000')
|
||||
if '2band_48000' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/2band_48000.json')
|
||||
param_name_auto=str('2band_48000')
|
||||
|
||||
#3 Band
|
||||
if '3band_44100' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100.json')
|
||||
param_name_auto=str('3band_44100')
|
||||
if '3band_44100_mid' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100_mid.json')
|
||||
param_name_auto=str('3band_44100_mid')
|
||||
if '3band_44100_msb2' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/3band_44100_msb2.json')
|
||||
param_name_auto=str('3band_44100_msb2')
|
||||
|
||||
#4 Band
|
||||
if '4band_44100' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100.json')
|
||||
param_name_auto=str('4band_44100')
|
||||
if '4band_44100_mid' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100_mid.json')
|
||||
param_name_auto=str('4band_44100_mid')
|
||||
if '4band_44100_msb' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100_msb.json')
|
||||
param_name_auto=str('4band_44100_msb')
|
||||
if '4band_44100_msb2' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100_msb2.json')
|
||||
param_name_auto=str('4band_44100_msb2')
|
||||
if '4band_44100_reverse' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100_reverse.json')
|
||||
param_name_auto=str('4band_44100_reverse')
|
||||
if '4band_44100_sw' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_44100_sw.json')
|
||||
param_name_auto=str('4band_44100_sw')
|
||||
if '4band_v2' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_v2.json')
|
||||
param_name_auto=str('4band_v2')
|
||||
if '4band_v2_sn' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/4band_v2_sn.json')
|
||||
param_name_auto=str('4band_v2_sn')
|
||||
if 'tmodelparam' in ModelName:
|
||||
model_params_auto=str('lib_v5/modelparams/tmodelparam.json')
|
||||
param_name_auto=str('User Model Param Set')
|
||||
|
||||
text_widget.write(' Done!\n')
|
||||
|
||||
text_widget.write('Done!\n')
|
||||
|
||||
if data['ModelParams'] == 'Auto':
|
||||
param_name = param_name_auto
|
||||
model_params_d = model_params_auto
|
||||
model_hash = hashlib.md5(open(ModelName,'rb').read()).hexdigest()
|
||||
model_params = []
|
||||
model_params = lib_v5.filelist.provide_model_param_hash(model_hash)
|
||||
print(model_params)
|
||||
if model_params[0] == 'Not Found Using Hash':
|
||||
model_params = []
|
||||
model_params = lib_v5.filelist.provide_model_param_name(ModelName)
|
||||
if model_params[0] == 'Not Found Using Name':
|
||||
text_widget.write(base_text + f'Unable to set model parameters automatically with the selected model.\n')
|
||||
confirm = tk.messagebox.askyesno(title='Unrecognized Model Detected',
|
||||
message=f'\nThe application could not automatically set the model param for the selected model.\n\n' +
|
||||
f'Would you like to select the model param file for this model?\n\n')
|
||||
|
||||
if confirm:
|
||||
model_param_selection = filedialog.askopenfilename(initialdir='lib_v5/modelparams',
|
||||
title=f'Select Model Param',
|
||||
filetypes=[("Model Param", "*.json")])
|
||||
|
||||
model_param_file_path = str(model_param_selection)
|
||||
model_param_file = os.path.splitext(os.path.basename(model_param_file_path))[0] + '.json'
|
||||
model_params = [model_param_file_path, model_param_file]
|
||||
|
||||
with open(f"lib_v5/filelists/model_cache/vr_param_cache/{model_hash}.txt", 'w') as f:
|
||||
f.write(model_param_file)
|
||||
|
||||
|
||||
if model_params[0] == '':
|
||||
text_widget.write("\n" + base_text + f'Separation failed for the following audio file:\n')
|
||||
text_widget.write(base_text + f'"{os.path.basename(music_file)}"\n')
|
||||
text_widget.write(f'\nError Received:\n\n')
|
||||
text_widget.write(f'Model parameters are missing.\n\n')
|
||||
text_widget.write(f'Please check the following:\n')
|
||||
text_widget.write(f'1. Make sure the model is still present.\n')
|
||||
text_widget.write(f'2. If you are running a model that was not originally included in this package, \nplease append the modelparam name to the model name.\n')
|
||||
text_widget.write(f' - Example if using \"4band_v2.json\" modelparam: \"model_4band_v2.pth\"\n\n')
|
||||
text_widget.write(f'Please address this and try again.\n\n')
|
||||
text_widget.write(f'Time Elapsed: {time.strftime("%H:%M:%S", time.gmtime(int(time.perf_counter() - stime)))}')
|
||||
torch.cuda.empty_cache()
|
||||
progress_var.set(0)
|
||||
button_widget.configure(state=tk.NORMAL) # Enable Button
|
||||
return
|
||||
|
||||
else:
|
||||
pass
|
||||
else:
|
||||
text_widget.write(base_text + f'Model param not selected.\n')
|
||||
text_widget.write("\n" + base_text + f'Separation failed for the following audio file:\n')
|
||||
text_widget.write(base_text + f'"{os.path.basename(music_file)}"\n')
|
||||
text_widget.write(f'\nError Received:\n\n')
|
||||
text_widget.write(f'Model parameters are missing.\n\n')
|
||||
text_widget.write(f'Please check the following:\n')
|
||||
text_widget.write(f'1. Make sure the model is still present.\n')
|
||||
text_widget.write(f'2. If you are running a model that was not originally included in this package, \nplease append the modelparam name to the model name.\n')
|
||||
text_widget.write(f' - Example if using \"4band_v2.json\" modelparam: \"model_4band_v2.pth\"\n\n')
|
||||
text_widget.write(f'Please address this and try again.\n\n')
|
||||
text_widget.write(f'Time Elapsed: {time.strftime("%H:%M:%S", time.gmtime(int(time.perf_counter() - stime)))}')
|
||||
torch.cuda.empty_cache()
|
||||
progress_var.set(0)
|
||||
button_widget.configure(state=tk.NORMAL) # Enable Button
|
||||
return
|
||||
|
||||
else:
|
||||
param_name = str(data['ModelParams'])
|
||||
model_params_d = str('lib_v5/modelparams/' + data['ModelParams'])
|
||||
param = data['ModelParams']
|
||||
model_param_file_path = f'lib_v5/modelparams/{param}'
|
||||
model_params = [model_param_file_path, param]
|
||||
|
||||
try:
|
||||
print('Model Parameters:', model_params_d)
|
||||
text_widget.write(base_text + 'Loading assigned model parameters ' + '\"' + param_name + '\"... ')
|
||||
except Exception as e:
|
||||
traceback_text = ''.join(traceback.format_tb(e.__traceback__))
|
||||
errmessage = f'Traceback Error: "{traceback_text}"\n{type(e).__name__}: "{e}"\n'
|
||||
text_widget.write("\n" + base_text + f'Separation failed for the following audio file:\n')
|
||||
text_widget.write(base_text + f'"{os.path.basename(music_file)}"\n')
|
||||
text_widget.write(f'\nError Received:\n\n')
|
||||
text_widget.write(f'Model parameters are missing.\n\n')
|
||||
text_widget.write(f'Please check the following:\n')
|
||||
text_widget.write(f'1. Make sure the model is still present.\n')
|
||||
text_widget.write(f'2. If you are running a model that was not originally included in this package, \nplease append the modelparam name to the model name.\n')
|
||||
text_widget.write(f' - Example if using \"4band_v2.json\" modelparam: \"model_4band_v2.pth\"\n\n')
|
||||
text_widget.write(f'Please address this and try again.\n\n')
|
||||
text_widget.write(f'Time Elapsed: {time.strftime("%H:%M:%S", time.gmtime(int(time.perf_counter() - stime)))}')
|
||||
try:
|
||||
with open('errorlog.txt', 'w') as f:
|
||||
f.write(f'Last Error Received:\n\n' +
|
||||
f'Error Received while processing "{os.path.basename(music_file)}":\n' +
|
||||
f'Process Method: VR Architecture\n\n' +
|
||||
f'Model parameters are missing.\n\n' +
|
||||
f'Please check the following:\n' +
|
||||
f'1. Make sure the model is still present.\n' +
|
||||
f'2. If you are running a model that was not originally included in this package, please append the modelparam name to the model name.\n' +
|
||||
f' - Example if using \"4band_v2.json\" modelparam: \"model_4band_v2.pth\"\n\n' +
|
||||
f'Please address this and try again.\n\n' +
|
||||
f'Raw error details:\n\n' +
|
||||
errmessage + f'\nError Time Stamp: [{datetime.now().strftime("%Y-%m-%d %H:%M:%S")}]\n')
|
||||
except:
|
||||
pass
|
||||
torch.cuda.empty_cache()
|
||||
progress_var.set(0)
|
||||
button_widget.configure(state=tk.NORMAL) # Enable Button
|
||||
return
|
||||
|
||||
|
||||
mp = ModelParameters(model_params_d)
|
||||
text_widget.write(base_text + 'Loading assigned model parameters ' + '\"' + model_params[1] + '\"... ')
|
||||
mp = ModelParameters(model_params[0])
|
||||
text_widget.write('Done!\n')
|
||||
# -Instrumental-
|
||||
if os.path.isfile(data['instrumentalModel']):
|
||||
@ -726,10 +584,8 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
|
||||
|
||||
model_name = os.path.basename(data[f'{data["useModel"]}Model'])
|
||||
|
||||
mp = ModelParameters(model_params_d)
|
||||
|
||||
# -Go through the different steps of seperation-
|
||||
# -Go through the different steps of Separation-
|
||||
# Wave source
|
||||
text_widget.write(base_text + 'Loading audio source...')
|
||||
|
||||
@ -921,12 +777,12 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
text_widget.write(base_text + 'Loading Demucs model... ')
|
||||
update_progress(**progress_kwargs,
|
||||
step=0.95)
|
||||
path_d = Path('models/Demucs_Models')
|
||||
print('What Demucs model was chosen? ', demucs_model_set)
|
||||
path_d = Path('models/Demucs_Models/v3_repo')
|
||||
#print('What Demucs model was chosen? ', demucs_model_set)
|
||||
demucs = _gm(name=demucs_model_set, repo=path_d)
|
||||
text_widget.write('Done!\n')
|
||||
|
||||
print('segment: ', data['segment'])
|
||||
#print('segment: ', data['segment'])
|
||||
|
||||
if data['segment'] == 'None':
|
||||
segment = None
|
||||
@ -958,7 +814,7 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
if segment is not None:
|
||||
sub.segment = segment
|
||||
|
||||
print('segment port-process: ', segment)
|
||||
#print('segment port-process: ', segment)
|
||||
|
||||
demucs.cpu()
|
||||
demucs.eval()
|
||||
@ -1048,7 +904,7 @@ def main(window: tk.Wm, text_widget: tk.Text, button_widget: tk.Button, progress
|
||||
bin_image.tofile(f)
|
||||
|
||||
|
||||
text_widget.write(base_text + 'Completed Seperation!\n\n')
|
||||
text_widget.write(base_text + 'Completed Separation!\n\n')
|
||||
except Exception as e:
|
||||
traceback_text = ''.join(traceback.format_tb(e.__traceback__))
|
||||
message = f'Traceback Error: "{traceback_text}"\n{type(e).__name__}: "{e}"\n'
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
12
models.py
12
models.py
@ -7,8 +7,6 @@ import librosa
|
||||
dim_c = 4
|
||||
model_path = 'model'
|
||||
|
||||
#n_fft_scale = {'vocals-one':6144, 'vocals-two':7680,'*':2}
|
||||
|
||||
class Conv_TDF_net_trim(nn.Module):
|
||||
def __init__(self, device, n_fft_scale, dim_f, load, model_name, target_name,
|
||||
L, dim_t, hop=1024):
|
||||
@ -20,9 +18,9 @@ class Conv_TDF_net_trim(nn.Module):
|
||||
self.hop = hop
|
||||
self.n_bins = self.n_fft//2+1
|
||||
self.chunk_size = hop * (self.dim_t-1)
|
||||
self.window = torch.hann_window(window_length=self.n_fft, periodic=True).to(device)
|
||||
self.window = torch.hann_window(window_length=self.n_fft, periodic=False).to(device)
|
||||
self.target_name = target_name
|
||||
print(n_fft_scale)
|
||||
#print(n_fft_scale)
|
||||
out_c = dim_c*4 if target_name=='*' else dim_c
|
||||
self.freq_pad = torch.zeros([1, out_c, self.n_bins-self.dim_f, self.dim_t]).to(device)
|
||||
|
||||
@ -61,17 +59,21 @@ def istft(spec, hl):
|
||||
return wave
|
||||
|
||||
def spec_effects(wave, algorithm='Default', value=None):
|
||||
spec = [stft(wave[0],2048,1024),stft(wave[1],2048,1024)]
|
||||
doubleout = spec = [stft(wave[0],2048,1024),stft(wave[1],2048,1024)]
|
||||
if algorithm == 'Min_Mag':
|
||||
doubleout
|
||||
v_spec_m = np.where(np.abs(spec[1]) <= np.abs(spec[0]), spec[1], spec[0])
|
||||
wave = istft(v_spec_m,1024)
|
||||
elif algorithm == 'Max_Mag':
|
||||
doubleout
|
||||
v_spec_m = np.where(np.abs(spec[1]) >= np.abs(spec[0]), spec[1], spec[0])
|
||||
wave = istft(v_spec_m,1024)
|
||||
elif algorithm == 'Default':
|
||||
doubleout
|
||||
#wave = [istft(spec[0],1024),istft(spec[1],1024)]
|
||||
wave = (wave[1] * value) + (wave[0] * (1-value))
|
||||
elif algorithm == 'Invert_p':
|
||||
doubleout
|
||||
X_mag = np.abs(spec[0])
|
||||
y_mag = np.abs(spec[1])
|
||||
max_mag = np.where(X_mag >= y_mag, X_mag, y_mag)
|
||||
|
||||
Loading…
Reference in New Issue
Block a user