Spaces:
Sleeping
Sleeping
Christian J. Steinmetz
commited on
Commit
·
d254115
1
Parent(s):
e8eaf47
adding updated support for classifier models
Browse files- cfg/exp/5-5_cls.yaml +59 -0
- cfg/model/cls_panns_16k.yaml +15 -0
- cfg/model/cls_panns_44k.yaml +15 -0
- cfg/model/{classifier.yaml → cls_panns_pt.yaml} +2 -5
- cfg/model/cls_vggish.yaml +11 -0
- cfg/model/cls_wav2clip.yaml +11 -0
- cfg/model/cls_wav2vec2.yaml +11 -0
- remfx/{cnn14.py → classifier.py} +159 -16
- setup.py +2 -2
- train_all.sh +6 -0
cfg/exp/5-5_cls.yaml
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
defaults:
|
3 |
+
- override /model: demucs
|
4 |
+
- override /effects: all
|
5 |
+
seed: 12345
|
6 |
+
sample_rate: 48000
|
7 |
+
chunk_size: 262144 # 5.5s
|
8 |
+
logs_dir: "./logs"
|
9 |
+
render_files: True
|
10 |
+
render_root: "/scratch/EffectSet_cjs"
|
11 |
+
accelerator: "gpu"
|
12 |
+
log_audio: False
|
13 |
+
# Effects
|
14 |
+
num_kept_effects: [0,0] # [min, max]
|
15 |
+
num_removed_effects: [0,5] # [min, max]
|
16 |
+
shuffle_kept_effects: True
|
17 |
+
shuffle_removed_effects: True
|
18 |
+
num_classes: 5
|
19 |
+
effects_to_keep:
|
20 |
+
effects_to_remove:
|
21 |
+
- distortion
|
22 |
+
- compressor
|
23 |
+
- reverb
|
24 |
+
- chorus
|
25 |
+
- delay
|
26 |
+
datamodule:
|
27 |
+
batch_size: 64
|
28 |
+
num_workers: 8
|
29 |
+
|
30 |
+
callbacks:
|
31 |
+
model_checkpoint:
|
32 |
+
_target_: pytorch_lightning.callbacks.ModelCheckpoint
|
33 |
+
monitor: "valid_f1_avg_epoch" # name of the logged metric which determines when model is improving
|
34 |
+
save_top_k: 1 # save k best models (determined by above metric)
|
35 |
+
save_last: True # additionaly always save model from last epoch
|
36 |
+
mode: "max" # can be "max" or "min"
|
37 |
+
verbose: True
|
38 |
+
dirpath: ${logs_dir}/ckpts/${now:%Y-%m-%d-%H-%M-%S}
|
39 |
+
filename: '{epoch:02d}-{valid_f1_avg_epoch:.3f}'
|
40 |
+
learning_rate_monitor:
|
41 |
+
_target_: pytorch_lightning.callbacks.LearningRateMonitor
|
42 |
+
logging_interval: "step"
|
43 |
+
#audio_logging:
|
44 |
+
# _target_: remfx.callbacks.AudioCallback
|
45 |
+
# sample_rate: ${sample_rate}
|
46 |
+
# log_audio: ${log_audio}
|
47 |
+
|
48 |
+
|
49 |
+
trainer:
|
50 |
+
_target_: pytorch_lightning.Trainer
|
51 |
+
precision: 32 # Precision used for tensors, default `32`
|
52 |
+
min_epochs: 0
|
53 |
+
max_epochs: -1
|
54 |
+
log_every_n_steps: 1 # Logs metrics every N batches
|
55 |
+
accumulate_grad_batches: 1
|
56 |
+
accelerator: ${accelerator}
|
57 |
+
devices: 1
|
58 |
+
gradient_clip_val: 10.0
|
59 |
+
max_steps: 150000
|
cfg/model/cls_panns_16k.yaml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
model:
|
3 |
+
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 3e-4
|
5 |
+
lr_weight_decay: 1e-3
|
6 |
+
sample_rate: ${sample_rate}
|
7 |
+
network:
|
8 |
+
_target_: remfx.classifier.Cnn14
|
9 |
+
num_classes: ${num_classes}
|
10 |
+
n_fft: 2048
|
11 |
+
hop_length: 512
|
12 |
+
n_mels: 128
|
13 |
+
sample_rate: 44100
|
14 |
+
model_sample_rate: 16000
|
15 |
+
|
cfg/model/cls_panns_44k.yaml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
model:
|
3 |
+
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 3e-4
|
5 |
+
lr_weight_decay: 1e-3
|
6 |
+
sample_rate: ${sample_rate}
|
7 |
+
network:
|
8 |
+
_target_: remfx.classifier.Cnn14
|
9 |
+
num_classes: ${num_classes}
|
10 |
+
n_fft: 1024
|
11 |
+
hop_length: 256
|
12 |
+
n_mels: 128
|
13 |
+
sample_rate: 44100
|
14 |
+
model_sample_rate: 44100
|
15 |
+
specaugment: True
|
cfg/model/{classifier.yaml → cls_panns_pt.yaml}
RENAMED
@@ -1,14 +1,11 @@
|
|
1 |
# @package _global_
|
2 |
model:
|
3 |
_target_: remfx.models.FXClassifier
|
4 |
-
lr:
|
5 |
lr_weight_decay: 1e-3
|
6 |
sample_rate: ${sample_rate}
|
7 |
network:
|
8 |
-
_target_: remfx.
|
9 |
num_classes: ${num_classes}
|
10 |
-
n_fft: 4096
|
11 |
-
hop_length: 512
|
12 |
-
n_mels: 128
|
13 |
sample_rate: ${sample_rate}
|
14 |
|
|
|
1 |
# @package _global_
|
2 |
model:
|
3 |
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 3e-4
|
5 |
lr_weight_decay: 1e-3
|
6 |
sample_rate: ${sample_rate}
|
7 |
network:
|
8 |
+
_target_: remfx.classifier.PANNs
|
9 |
num_classes: ${num_classes}
|
|
|
|
|
|
|
10 |
sample_rate: ${sample_rate}
|
11 |
|
cfg/model/cls_vggish.yaml
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
model:
|
3 |
+
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 3e-4
|
5 |
+
lr_weight_decay: 1e-3
|
6 |
+
sample_rate: ${sample_rate}
|
7 |
+
network:
|
8 |
+
_target_: remfx.classifier.VGGish
|
9 |
+
num_classes: ${num_classes}
|
10 |
+
sample_rate: ${sample_rate}
|
11 |
+
|
cfg/model/cls_wav2clip.yaml
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
model:
|
3 |
+
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 3e-4
|
5 |
+
lr_weight_decay: 1e-3
|
6 |
+
sample_rate: ${sample_rate}
|
7 |
+
network:
|
8 |
+
_target_: remfx.classifier.Wav2CLIP
|
9 |
+
num_classes: ${num_classes}
|
10 |
+
sample_rate: ${sample_rate}
|
11 |
+
|
cfg/model/cls_wav2vec2.yaml
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
model:
|
3 |
+
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 3e-4
|
5 |
+
lr_weight_decay: 1e-3
|
6 |
+
sample_rate: ${sample_rate}
|
7 |
+
network:
|
8 |
+
_target_: remfx.classifier.wav2vec2
|
9 |
+
num_classes: ${num_classes}
|
10 |
+
sample_rate: ${sample_rate}
|
11 |
+
|
remfx/{cnn14.py → classifier.py}
RENAMED
@@ -1,8 +1,132 @@
|
|
1 |
import torch
|
2 |
import torchaudio
|
3 |
import torch.nn as nn
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
import torch.nn.functional as F
|
5 |
-
from utils import init_bn, init_layer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
|
7 |
# adapted from https://github.com/qiuqiangkong/audioset_tagging_cnn/blob/master/pytorch/models.py
|
8 |
|
@@ -12,20 +136,25 @@ class Cnn14(nn.Module):
|
|
12 |
self,
|
13 |
num_classes: int,
|
14 |
sample_rate: float,
|
15 |
-
|
16 |
-
|
|
|
17 |
n_mels: int = 128,
|
|
|
18 |
):
|
19 |
super().__init__()
|
20 |
self.num_classes = num_classes
|
21 |
self.n_fft = n_fft
|
22 |
self.hop_length = hop_length
|
|
|
|
|
|
|
23 |
|
24 |
window = torch.hann_window(n_fft)
|
25 |
self.register_buffer("window", window)
|
26 |
|
27 |
self.melspec = torchaudio.transforms.MelSpectrogram(
|
28 |
-
|
29 |
n_fft,
|
30 |
hop_length=hop_length,
|
31 |
n_mels=n_mels,
|
@@ -45,42 +174,56 @@ class Cnn14(nn.Module):
|
|
45 |
|
46 |
self.init_weight()
|
47 |
|
|
|
|
|
|
|
|
|
|
|
48 |
def init_weight(self):
|
49 |
init_bn(self.bn0)
|
50 |
init_layer(self.fc1)
|
51 |
init_layer(self.fc_audioset)
|
52 |
|
53 |
-
def forward(self, x: torch.Tensor):
|
54 |
"""
|
55 |
Input: (batch_size, data_length)"""
|
56 |
|
|
|
|
|
|
|
57 |
x = self.melspec(x)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
x = x.permute(0, 2, 1, 3)
|
59 |
x = self.bn0(x)
|
60 |
x = x.permute(0, 2, 1, 3)
|
61 |
|
62 |
-
if self.training:
|
63 |
-
pass
|
64 |
-
# x = self.spec_augmenter(x)
|
65 |
-
|
66 |
x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
|
67 |
-
x = F.dropout(x, p=0.2, training=
|
68 |
x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
|
69 |
-
x = F.dropout(x, p=0.2, training=
|
70 |
x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
|
71 |
-
x = F.dropout(x, p=0.2, training=
|
72 |
x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
|
73 |
-
x = F.dropout(x, p=0.2, training=
|
74 |
x = self.conv_block5(x, pool_size=(2, 2), pool_type="avg")
|
75 |
-
x = F.dropout(x, p=0.2, training=
|
76 |
x = self.conv_block6(x, pool_size=(1, 1), pool_type="avg")
|
77 |
-
x = F.dropout(x, p=0.2, training=
|
78 |
x = torch.mean(x, dim=3)
|
79 |
|
80 |
(x1, _) = torch.max(x, dim=2)
|
81 |
x2 = torch.mean(x, dim=2)
|
82 |
x = x1 + x2
|
83 |
-
x = F.dropout(x, p=0.5, training=
|
84 |
x = F.relu_(self.fc1(x))
|
85 |
clipwise_output = self.fc_audioset(x)
|
86 |
|
|
|
1 |
import torch
|
2 |
import torchaudio
|
3 |
import torch.nn as nn
|
4 |
+
import hearbaseline
|
5 |
+
import hearbaseline.vggish
|
6 |
+
import hearbaseline.wav2vec2
|
7 |
+
|
8 |
+
import wav2clip_hear
|
9 |
+
import panns_hear
|
10 |
+
|
11 |
+
|
12 |
import torch.nn.functional as F
|
13 |
+
from remfx.utils import init_bn, init_layer
|
14 |
+
|
15 |
+
|
16 |
+
class PANNs(torch.nn.Module):
|
17 |
+
def __init__(
|
18 |
+
self, num_classes: int, sample_rate: float, hidden_dim: int = 256
|
19 |
+
) -> None:
|
20 |
+
super().__init__()
|
21 |
+
self.num_classes = num_classes
|
22 |
+
self.model = panns_hear.load_model("hear2021-panns_hear.pth")
|
23 |
+
self.resample = torchaudio.transforms.Resample(
|
24 |
+
orig_freq=sample_rate, new_freq=32000
|
25 |
+
)
|
26 |
+
self.proj = torch.nn.Sequential(
|
27 |
+
torch.nn.Linear(2048, hidden_dim),
|
28 |
+
torch.nn.ReLU(),
|
29 |
+
torch.nn.Linear(hidden_dim, hidden_dim),
|
30 |
+
torch.nn.ReLU(),
|
31 |
+
torch.nn.Linear(hidden_dim, num_classes),
|
32 |
+
)
|
33 |
+
|
34 |
+
def forward(self, x: torch.Tensor):
|
35 |
+
with torch.no_grad():
|
36 |
+
x = self.resample(x)
|
37 |
+
embed = panns_hear.get_scene_embeddings(x.view(x.shape[0], -1), self.model)
|
38 |
+
return self.proj(embed)
|
39 |
+
|
40 |
+
|
41 |
+
class Wav2CLIP(nn.Module):
|
42 |
+
def __init__(
|
43 |
+
self,
|
44 |
+
num_classes: int,
|
45 |
+
sample_rate: float,
|
46 |
+
hidden_dim: int = 256,
|
47 |
+
) -> None:
|
48 |
+
super().__init__()
|
49 |
+
self.num_classes = num_classes
|
50 |
+
self.model = wav2clip_hear.load_model("")
|
51 |
+
self.resample = torchaudio.transforms.Resample(
|
52 |
+
orig_freq=sample_rate, new_freq=16000
|
53 |
+
)
|
54 |
+
self.proj = torch.nn.Sequential(
|
55 |
+
torch.nn.Linear(512, hidden_dim),
|
56 |
+
torch.nn.ReLU(),
|
57 |
+
torch.nn.Linear(hidden_dim, hidden_dim),
|
58 |
+
torch.nn.ReLU(),
|
59 |
+
torch.nn.Linear(hidden_dim, num_classes),
|
60 |
+
)
|
61 |
+
|
62 |
+
def forward(self, x: torch.Tensor):
|
63 |
+
with torch.no_grad():
|
64 |
+
x = self.resample(x)
|
65 |
+
embed = wav2clip_hear.get_scene_embeddings(
|
66 |
+
x.view(x.shape[0], -1), self.model
|
67 |
+
)
|
68 |
+
return self.proj(embed)
|
69 |
+
|
70 |
+
|
71 |
+
class VGGish(nn.Module):
|
72 |
+
def __init__(
|
73 |
+
self,
|
74 |
+
num_classes: int,
|
75 |
+
sample_rate: float,
|
76 |
+
hidden_dim: int = 256,
|
77 |
+
):
|
78 |
+
super().__init__()
|
79 |
+
self.num_classes = num_classes
|
80 |
+
self.resample = torchaudio.transforms.Resample(
|
81 |
+
orig_freq=sample_rate, new_freq=16000
|
82 |
+
)
|
83 |
+
self.model = hearbaseline.vggish.load_model()
|
84 |
+
self.proj = torch.nn.Sequential(
|
85 |
+
torch.nn.Linear(128, hidden_dim),
|
86 |
+
torch.nn.ReLU(),
|
87 |
+
torch.nn.Linear(hidden_dim, hidden_dim),
|
88 |
+
torch.nn.ReLU(),
|
89 |
+
torch.nn.Linear(hidden_dim, num_classes),
|
90 |
+
)
|
91 |
+
|
92 |
+
def forward(self, x: torch.Tensor):
|
93 |
+
with torch.no_grad():
|
94 |
+
x = self.resample(x)
|
95 |
+
embed = hearbaseline.vggish.get_scene_embeddings(
|
96 |
+
x.view(x.shape[0], -1), self.model
|
97 |
+
)
|
98 |
+
return self.proj(embed)
|
99 |
+
|
100 |
+
|
101 |
+
class wav2vec2(nn.Module):
|
102 |
+
def __init__(
|
103 |
+
self,
|
104 |
+
num_classes: int,
|
105 |
+
sample_rate: float,
|
106 |
+
hidden_dim: int = 256,
|
107 |
+
):
|
108 |
+
super().__init__()
|
109 |
+
self.num_classes = num_classes
|
110 |
+
self.resample = torchaudio.transforms.Resample(
|
111 |
+
orig_freq=sample_rate, new_freq=16000
|
112 |
+
)
|
113 |
+
self.model = hearbaseline.wav2vec2.load_model()
|
114 |
+
self.proj = torch.nn.Sequential(
|
115 |
+
torch.nn.Linear(1024, hidden_dim),
|
116 |
+
torch.nn.ReLU(),
|
117 |
+
torch.nn.Linear(hidden_dim, hidden_dim),
|
118 |
+
torch.nn.ReLU(),
|
119 |
+
torch.nn.Linear(hidden_dim, num_classes),
|
120 |
+
)
|
121 |
+
|
122 |
+
def forward(self, x: torch.Tensor):
|
123 |
+
with torch.no_grad():
|
124 |
+
x = self.resample(x)
|
125 |
+
embed = hearbaseline.wav2vec2.get_scene_embeddings(
|
126 |
+
x.view(x.shape[0], -1), self.model
|
127 |
+
)
|
128 |
+
return self.proj(embed)
|
129 |
+
|
130 |
|
131 |
# adapted from https://github.com/qiuqiangkong/audioset_tagging_cnn/blob/master/pytorch/models.py
|
132 |
|
|
|
136 |
self,
|
137 |
num_classes: int,
|
138 |
sample_rate: float,
|
139 |
+
model_sample_rate: float,
|
140 |
+
n_fft: int = 1024,
|
141 |
+
hop_length: int = 256,
|
142 |
n_mels: int = 128,
|
143 |
+
specaugment: bool = False,
|
144 |
):
|
145 |
super().__init__()
|
146 |
self.num_classes = num_classes
|
147 |
self.n_fft = n_fft
|
148 |
self.hop_length = hop_length
|
149 |
+
self.sample_rate = sample_rate
|
150 |
+
self.model_sample_rate = model_sample_rate
|
151 |
+
self.specaugment = specaugment
|
152 |
|
153 |
window = torch.hann_window(n_fft)
|
154 |
self.register_buffer("window", window)
|
155 |
|
156 |
self.melspec = torchaudio.transforms.MelSpectrogram(
|
157 |
+
model_sample_rate,
|
158 |
n_fft,
|
159 |
hop_length=hop_length,
|
160 |
n_mels=n_mels,
|
|
|
174 |
|
175 |
self.init_weight()
|
176 |
|
177 |
+
if sample_rate != model_sample_rate:
|
178 |
+
self.resample = torchaudio.transforms.Resample(
|
179 |
+
orig_freq=sample_rate, new_freq=model_sample_rate
|
180 |
+
)
|
181 |
+
|
182 |
def init_weight(self):
|
183 |
init_bn(self.bn0)
|
184 |
init_layer(self.fc1)
|
185 |
init_layer(self.fc_audioset)
|
186 |
|
187 |
+
def forward(self, x: torch.Tensor, train: bool = False):
|
188 |
"""
|
189 |
Input: (batch_size, data_length)"""
|
190 |
|
191 |
+
if self.sample_rate != self.model_sample_rate:
|
192 |
+
x = self.resample(x)
|
193 |
+
|
194 |
x = self.melspec(x)
|
195 |
+
|
196 |
+
if self.specaugment and train:
|
197 |
+
# import matplotlib.pyplot as plt
|
198 |
+
# fig, axs = plt.subplots(2, 1, sharex=True)
|
199 |
+
# axs[0].imshow(x[0, :, :, :].detach().squeeze().cpu().numpy())
|
200 |
+
x = self.freq_mask(x)
|
201 |
+
x = self.time_mask(x)
|
202 |
+
# axs[1].imshow(x[0, :, :, :].detach().squeeze().cpu().numpy())
|
203 |
+
# plt.savefig("spec_augment.png", dpi=300)
|
204 |
+
|
205 |
x = x.permute(0, 2, 1, 3)
|
206 |
x = self.bn0(x)
|
207 |
x = x.permute(0, 2, 1, 3)
|
208 |
|
|
|
|
|
|
|
|
|
209 |
x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
|
210 |
+
x = F.dropout(x, p=0.2, training=train)
|
211 |
x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
|
212 |
+
x = F.dropout(x, p=0.2, training=train)
|
213 |
x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
|
214 |
+
x = F.dropout(x, p=0.2, training=train)
|
215 |
x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
|
216 |
+
x = F.dropout(x, p=0.2, training=train)
|
217 |
x = self.conv_block5(x, pool_size=(2, 2), pool_type="avg")
|
218 |
+
x = F.dropout(x, p=0.2, training=train)
|
219 |
x = self.conv_block6(x, pool_size=(1, 1), pool_type="avg")
|
220 |
+
x = F.dropout(x, p=0.2, training=train)
|
221 |
x = torch.mean(x, dim=3)
|
222 |
|
223 |
(x1, _) = torch.max(x, dim=2)
|
224 |
x2 = torch.mean(x, dim=2)
|
225 |
x = x1 + x2
|
226 |
+
x = F.dropout(x, p=0.5, training=train)
|
227 |
x = F.relu_(self.fc1(x))
|
228 |
clipwise_output = self.fc_audioset(x)
|
229 |
|
setup.py
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
from pathlib import Path
|
2 |
from setuptools import setup, find_packages
|
3 |
|
4 |
-
NAME = "
|
5 |
-
DESCRIPTION = ""
|
6 |
URL = ""
|
7 |
EMAIL = "[email protected]"
|
8 |
AUTHOR = "Matthew Rice"
|
|
|
1 |
from pathlib import Path
|
2 |
from setuptools import setup, find_packages
|
3 |
|
4 |
+
NAME = "remfx"
|
5 |
+
DESCRIPTION = "Universal audio effect removal"
|
6 |
URL = ""
|
7 |
EMAIL = "[email protected]"
|
8 |
AUTHOR = "Matthew Rice"
|
train_all.sh
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
python scripts/train.py +exp=5-5_cls.yaml model=cls_wav2vec2 render_files=False logs_dir=/scratch/cjs-log
|
2 |
+
python scripts/train.py +exp=5-5_cls.yaml model=cls_panns_44k render_files=False logs_dir=/scratch/cjs-log
|
3 |
+
python scripts/train.py +exp=5-5_cls.yaml model=cls_panns_16k render_files=False logs_dir=/scratch/cjs-log
|
4 |
+
python scripts/train.py +exp=5-5_cls.yaml model=cls_panns_pt render_files=False logs_dir=/scratch/cjs-log
|
5 |
+
python scripts/train.py +exp=5-5_cls.yaml model=cls_vggish render_files=False logs_dir=/scratch/cjs-log
|
6 |
+
python scripts/train.py +exp=5-5_cls.yaml model=cls_wav2clip render_files=False logs_dir=/scratch/cjs-log
|