Spaces:
Runtime error
Runtime error
mattricesound
commited on
Merge pull request #29 from mhrice/cjs--classifier-v2
Browse files- README.md +1 -2
- cfg/config.yaml +1 -0
- cfg/model/classifier.yaml +14 -0
- remfx/datasets.py +71 -26
- remfx/effects.py +3 -3
- remfx/models.py +229 -13
README.md
CHANGED
@@ -9,10 +9,9 @@
|
|
9 |
5. `pip install -e umx`
|
10 |
|
11 |
## Download [VocalSet Dataset](https://zenodo.org/record/1193957)
|
12 |
-
1. `wget https://zenodo.org/record/
|
13 |
2. `mv VocalSet.zip?download=1 VocalSet.zip`
|
14 |
3. `unzip VocalSet.zip`
|
15 |
-
4. Manually split singers into train, val, test directories
|
16 |
|
17 |
# Training
|
18 |
## Steps
|
|
|
9 |
5. `pip install -e umx`
|
10 |
|
11 |
## Download [VocalSet Dataset](https://zenodo.org/record/1193957)
|
12 |
+
1. `wget https://zenodo.org/record/1442513/files/VocalSet1-2.zip?download=1`
|
13 |
2. `mv VocalSet.zip?download=1 VocalSet.zip`
|
14 |
3. `unzip VocalSet.zip`
|
|
|
15 |
|
16 |
# Training
|
17 |
## Steps
|
cfg/config.yaml
CHANGED
@@ -16,6 +16,7 @@ max_kept_effects: -1
|
|
16 |
max_removed_effects: -1
|
17 |
shuffle_kept_effects: True
|
18 |
shuffle_removed_effects: False
|
|
|
19 |
effects_to_use:
|
20 |
- compressor
|
21 |
- distortion
|
|
|
16 |
max_removed_effects: -1
|
17 |
shuffle_kept_effects: True
|
18 |
shuffle_removed_effects: False
|
19 |
+
num_classes: 4
|
20 |
effects_to_use:
|
21 |
- compressor
|
22 |
- distortion
|
cfg/model/classifier.yaml
ADDED
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# @package _global_
|
2 |
+
model:
|
3 |
+
_target_: remfx.models.FXClassifier
|
4 |
+
lr: 1e-4
|
5 |
+
lr_weight_decay: 1e-3
|
6 |
+
sample_rate: ${sample_rate}
|
7 |
+
network:
|
8 |
+
_target_: remfx.models.Cnn14
|
9 |
+
num_classes: ${num_classes}
|
10 |
+
n_fft: 4096
|
11 |
+
hop_length: 512
|
12 |
+
n_mels: 128
|
13 |
+
sample_rate: ${sample_rate}
|
14 |
+
|
remfx/datasets.py
CHANGED
@@ -1,16 +1,19 @@
|
|
|
|
|
|
|
|
1 |
import torch
|
2 |
-
|
3 |
-
import torch.nn.functional as F
|
4 |
import torchaudio
|
5 |
-
from pathlib import Path
|
6 |
import pytorch_lightning as pl
|
7 |
-
import
|
8 |
-
|
9 |
-
from remfx import effects
|
10 |
from tqdm import tqdm
|
11 |
-
from
|
12 |
-
import
|
13 |
from ordered_set import OrderedSet
|
|
|
|
|
|
|
14 |
|
15 |
|
16 |
# https://zenodo.org/record/1193957 -> VocalSet
|
@@ -18,6 +21,30 @@ from ordered_set import OrderedSet
|
|
18 |
ALL_EFFECTS = effects.Pedalboard_Effects
|
19 |
|
20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
class VocalSet(Dataset):
|
22 |
def __init__(
|
23 |
self,
|
@@ -43,8 +70,6 @@ class VocalSet(Dataset):
|
|
43 |
self.chunk_size = chunk_size
|
44 |
self.sample_rate = sample_rate
|
45 |
self.mode = mode
|
46 |
-
mode_path = self.root / self.mode
|
47 |
-
self.files = sorted(list(mode_path.glob("./**/*.wav")))
|
48 |
self.max_kept_effects = max_kept_effects
|
49 |
self.max_removed_effects = max_removed_effects
|
50 |
self.effects_to_use = effects_to_use
|
@@ -53,11 +78,20 @@ class VocalSet(Dataset):
|
|
53 |
self.effects = effect_modules
|
54 |
self.shuffle_kept_effects = shuffle_kept_effects
|
55 |
self.shuffle_removed_effects = shuffle_removed_effects
|
56 |
-
|
57 |
effects_string = "_".join(self.effects_to_use + ["_"] + self.effects_to_remove)
|
58 |
self.effects_to_keep = self.validate_effect_input()
|
59 |
self.proc_root = self.render_root / "processed" / effects_string / self.mode
|
60 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
if self.proc_root.exists() and len(list(self.proc_root.iterdir())) > 0:
|
62 |
print("Found processed files.")
|
63 |
if render_files:
|
@@ -86,12 +120,15 @@ class VocalSet(Dataset):
|
|
86 |
# Skip if chunk is too small
|
87 |
continue
|
88 |
|
89 |
-
dry, wet,
|
|
|
|
|
90 |
output_dir = self.proc_root / str(self.num_chunks)
|
91 |
output_dir.mkdir(exist_ok=True)
|
92 |
torchaudio.save(output_dir / "input.wav", wet, self.sample_rate)
|
93 |
torchaudio.save(output_dir / "target.wav", dry, self.sample_rate)
|
94 |
-
torch.save(
|
|
|
95 |
self.num_chunks += 1
|
96 |
else:
|
97 |
self.num_chunks = len(list(self.proc_root.iterdir()))
|
@@ -107,10 +144,11 @@ class VocalSet(Dataset):
|
|
107 |
def __getitem__(self, idx):
|
108 |
input_file = self.proc_root / str(idx) / "input.wav"
|
109 |
target_file = self.proc_root / str(idx) / "target.wav"
|
110 |
-
|
|
|
111 |
input, sr = torchaudio.load(input_file)
|
112 |
target, sr = torchaudio.load(target_file)
|
113 |
-
return (input, target,
|
114 |
|
115 |
def validate_effect_input(self):
|
116 |
for effect in self.effects.values():
|
@@ -154,27 +192,28 @@ class VocalSet(Dataset):
|
|
154 |
return kept_fx
|
155 |
|
156 |
def process_effects(self, dry: torch.Tensor):
|
157 |
-
labels = []
|
158 |
-
|
159 |
# Apply Kept Effects
|
160 |
# Shuffle effects if specified
|
161 |
if self.shuffle_kept_effects:
|
162 |
effect_indices = torch.randperm(len(self.effects_to_keep))
|
163 |
else:
|
164 |
effect_indices = torch.arange(len(self.effects_to_keep))
|
|
|
165 |
# Up to max_kept_effects
|
166 |
if self.max_kept_effects != -1:
|
167 |
num_kept_effects = int(torch.rand(1).item() * (self.max_kept_effects)) + 1
|
168 |
else:
|
169 |
num_kept_effects = len(self.effects_to_keep)
|
170 |
effect_indices = effect_indices[:num_kept_effects]
|
|
|
171 |
# Index in effect settings
|
172 |
effect_names_to_apply = [self.effects_to_keep[i] for i in effect_indices]
|
173 |
effects_to_apply = [self.effects[i] for i in effect_names_to_apply]
|
174 |
# Apply
|
|
|
175 |
for effect in effects_to_apply:
|
176 |
dry = effect(dry)
|
177 |
-
|
178 |
|
179 |
# Apply effects_to_remove
|
180 |
# Shuffle effects if specified
|
@@ -185,9 +224,7 @@ class VocalSet(Dataset):
|
|
185 |
effect_indices = torch.arange(len(self.effects_to_remove))
|
186 |
# Up to max_removed_effects
|
187 |
if self.max_removed_effects != -1:
|
188 |
-
num_kept_effects = (
|
189 |
-
int(torch.rand(1).item() * (self.max_removed_effects)) + 1
|
190 |
-
)
|
191 |
else:
|
192 |
num_kept_effects = len(self.effects_to_remove)
|
193 |
effect_indices = effect_indices[: self.max_removed_effects]
|
@@ -195,17 +232,25 @@ class VocalSet(Dataset):
|
|
195 |
effect_names_to_apply = [self.effects_to_remove[i] for i in effect_indices]
|
196 |
effects_to_apply = [self.effects[i] for i in effect_names_to_apply]
|
197 |
# Apply
|
|
|
|
|
198 |
for effect in effects_to_apply:
|
199 |
wet = effect(wet)
|
200 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
201 |
|
202 |
-
# Convert labels to one-hot
|
203 |
-
one_hot = F.one_hot(torch.tensor(labels), num_classes=len(ALL_EFFECTS))
|
204 |
-
effects_present = torch.sum(one_hot, dim=0).float()
|
205 |
# Normalize
|
206 |
normalized_dry = self.normalize(dry)
|
207 |
normalized_wet = self.normalize(wet)
|
208 |
-
return normalized_dry, normalized_wet,
|
209 |
|
210 |
|
211 |
class VocalSetDatamodule(pl.LightningDataModule):
|
|
|
1 |
+
import os
|
2 |
+
import sys
|
3 |
+
import glob
|
4 |
import torch
|
5 |
+
import shutil
|
|
|
6 |
import torchaudio
|
|
|
7 |
import pytorch_lightning as pl
|
8 |
+
import torch.nn.functional as F
|
9 |
+
|
|
|
10 |
from tqdm import tqdm
|
11 |
+
from pathlib import Path
|
12 |
+
from remfx import effects
|
13 |
from ordered_set import OrderedSet
|
14 |
+
from typing import Any, List, Dict
|
15 |
+
from torch.utils.data import Dataset, DataLoader
|
16 |
+
from remfx.utils import create_sequential_chunks
|
17 |
|
18 |
|
19 |
# https://zenodo.org/record/1193957 -> VocalSet
|
|
|
21 |
ALL_EFFECTS = effects.Pedalboard_Effects
|
22 |
|
23 |
|
24 |
+
singer_splits = {
|
25 |
+
"train": [
|
26 |
+
"male1",
|
27 |
+
"male2",
|
28 |
+
"male3",
|
29 |
+
"male4",
|
30 |
+
"male5",
|
31 |
+
"male6",
|
32 |
+
"male7",
|
33 |
+
"male8",
|
34 |
+
"male9",
|
35 |
+
"female1",
|
36 |
+
"female2",
|
37 |
+
"female3",
|
38 |
+
"female4",
|
39 |
+
"female5",
|
40 |
+
"female6",
|
41 |
+
"female7",
|
42 |
+
],
|
43 |
+
"val": ["male10", "female8"],
|
44 |
+
"test": ["male11", "female9"],
|
45 |
+
}
|
46 |
+
|
47 |
+
|
48 |
class VocalSet(Dataset):
|
49 |
def __init__(
|
50 |
self,
|
|
|
70 |
self.chunk_size = chunk_size
|
71 |
self.sample_rate = sample_rate
|
72 |
self.mode = mode
|
|
|
|
|
73 |
self.max_kept_effects = max_kept_effects
|
74 |
self.max_removed_effects = max_removed_effects
|
75 |
self.effects_to_use = effects_to_use
|
|
|
78 |
self.effects = effect_modules
|
79 |
self.shuffle_kept_effects = shuffle_kept_effects
|
80 |
self.shuffle_removed_effects = shuffle_removed_effects
|
|
|
81 |
effects_string = "_".join(self.effects_to_use + ["_"] + self.effects_to_remove)
|
82 |
self.effects_to_keep = self.validate_effect_input()
|
83 |
self.proc_root = self.render_root / "processed" / effects_string / self.mode
|
84 |
|
85 |
+
# find all singer directories
|
86 |
+
singer_dirs = glob.glob(os.path.join(self.root, "data_by_singer", "*"))
|
87 |
+
singer_dirs = [
|
88 |
+
sd for sd in singer_dirs if os.path.basename(sd) in singer_splits[mode]
|
89 |
+
]
|
90 |
+
self.files = []
|
91 |
+
for singer_dir in singer_dirs:
|
92 |
+
self.files += glob.glob(os.path.join(singer_dir, "**", "**", "*.wav"))
|
93 |
+
self.files = sorted(self.files)
|
94 |
+
|
95 |
if self.proc_root.exists() and len(list(self.proc_root.iterdir())) > 0:
|
96 |
print("Found processed files.")
|
97 |
if render_files:
|
|
|
120 |
# Skip if chunk is too small
|
121 |
continue
|
122 |
|
123 |
+
dry, wet, dry_effects, wet_effects = self.process_effects(
|
124 |
+
resampled_chunk
|
125 |
+
)
|
126 |
output_dir = self.proc_root / str(self.num_chunks)
|
127 |
output_dir.mkdir(exist_ok=True)
|
128 |
torchaudio.save(output_dir / "input.wav", wet, self.sample_rate)
|
129 |
torchaudio.save(output_dir / "target.wav", dry, self.sample_rate)
|
130 |
+
torch.save(dry_effects, output_dir / "dry_effects.pt")
|
131 |
+
torch.save(wet_effects, output_dir / "wet_effects.pt")
|
132 |
self.num_chunks += 1
|
133 |
else:
|
134 |
self.num_chunks = len(list(self.proc_root.iterdir()))
|
|
|
144 |
def __getitem__(self, idx):
|
145 |
input_file = self.proc_root / str(idx) / "input.wav"
|
146 |
target_file = self.proc_root / str(idx) / "target.wav"
|
147 |
+
dry_effect_names = torch.load(self.proc_root / str(idx) / "dry_effects.pt")
|
148 |
+
wet_effect_names = torch.load(self.proc_root / str(idx) / "wet_effects.pt")
|
149 |
input, sr = torchaudio.load(input_file)
|
150 |
target, sr = torchaudio.load(target_file)
|
151 |
+
return (input, target, dry_effect_names, wet_effect_names)
|
152 |
|
153 |
def validate_effect_input(self):
|
154 |
for effect in self.effects.values():
|
|
|
192 |
return kept_fx
|
193 |
|
194 |
def process_effects(self, dry: torch.Tensor):
|
|
|
|
|
195 |
# Apply Kept Effects
|
196 |
# Shuffle effects if specified
|
197 |
if self.shuffle_kept_effects:
|
198 |
effect_indices = torch.randperm(len(self.effects_to_keep))
|
199 |
else:
|
200 |
effect_indices = torch.arange(len(self.effects_to_keep))
|
201 |
+
|
202 |
# Up to max_kept_effects
|
203 |
if self.max_kept_effects != -1:
|
204 |
num_kept_effects = int(torch.rand(1).item() * (self.max_kept_effects)) + 1
|
205 |
else:
|
206 |
num_kept_effects = len(self.effects_to_keep)
|
207 |
effect_indices = effect_indices[:num_kept_effects]
|
208 |
+
|
209 |
# Index in effect settings
|
210 |
effect_names_to_apply = [self.effects_to_keep[i] for i in effect_indices]
|
211 |
effects_to_apply = [self.effects[i] for i in effect_names_to_apply]
|
212 |
# Apply
|
213 |
+
dry_labels = []
|
214 |
for effect in effects_to_apply:
|
215 |
dry = effect(dry)
|
216 |
+
dry_labels.append(ALL_EFFECTS.index(type(effect)))
|
217 |
|
218 |
# Apply effects_to_remove
|
219 |
# Shuffle effects if specified
|
|
|
224 |
effect_indices = torch.arange(len(self.effects_to_remove))
|
225 |
# Up to max_removed_effects
|
226 |
if self.max_removed_effects != -1:
|
227 |
+
num_kept_effects = int(torch.rand(1).item() * (self.max_removed_effects))
|
|
|
|
|
228 |
else:
|
229 |
num_kept_effects = len(self.effects_to_remove)
|
230 |
effect_indices = effect_indices[: self.max_removed_effects]
|
|
|
232 |
effect_names_to_apply = [self.effects_to_remove[i] for i in effect_indices]
|
233 |
effects_to_apply = [self.effects[i] for i in effect_names_to_apply]
|
234 |
# Apply
|
235 |
+
|
236 |
+
wet_labels = []
|
237 |
for effect in effects_to_apply:
|
238 |
wet = effect(wet)
|
239 |
+
wet_labels.append(ALL_EFFECTS.index(type(effect)))
|
240 |
+
|
241 |
+
wet_labels_tensor = torch.zeros(len(ALL_EFFECTS))
|
242 |
+
dry_labels_tensor = torch.zeros(len(ALL_EFFECTS))
|
243 |
+
|
244 |
+
for label_idx in wet_labels:
|
245 |
+
wet_labels_tensor[label_idx] = 1.0
|
246 |
+
|
247 |
+
for label_idx in dry_labels:
|
248 |
+
dry_labels_tensor[label_idx] = 1.0
|
249 |
|
|
|
|
|
|
|
250 |
# Normalize
|
251 |
normalized_dry = self.normalize(dry)
|
252 |
normalized_wet = self.normalize(wet)
|
253 |
+
return normalized_dry, normalized_wet, dry_labels_tensor, wet_labels_tensor
|
254 |
|
255 |
|
256 |
class VocalSetDatamodule(pl.LightningDataModule):
|
remfx/effects.py
CHANGED
@@ -701,9 +701,9 @@ class RandomAudioEffectsChannel(torch.nn.Module):
|
|
701 |
Pedalboard_Effects = [
|
702 |
RandomPedalboardReverb,
|
703 |
RandomPedalboardChorus,
|
704 |
-
RandomPedalboardDelay,
|
705 |
RandomPedalboardDistortion,
|
706 |
RandomPedalboardCompressor,
|
707 |
-
RandomPedalboardPhaser,
|
708 |
-
RandomPedalboardLimiter,
|
709 |
]
|
|
|
701 |
Pedalboard_Effects = [
|
702 |
RandomPedalboardReverb,
|
703 |
RandomPedalboardChorus,
|
704 |
+
# RandomPedalboardDelay,
|
705 |
RandomPedalboardDistortion,
|
706 |
RandomPedalboardCompressor,
|
707 |
+
# RandomPedalboardPhaser,
|
708 |
+
# RandomPedalboardLimiter,
|
709 |
]
|
remfx/models.py
CHANGED
@@ -1,15 +1,19 @@
|
|
|
|
1 |
import torch
|
2 |
-
|
|
|
3 |
import pytorch_lightning as pl
|
|
|
|
|
|
|
4 |
from einops import rearrange
|
5 |
-
import
|
6 |
from audio_diffusion_pytorch import DiffusionModel
|
7 |
from auraloss.time import SISDRLoss
|
8 |
from auraloss.freq import MultiResolutionSTFTLoss
|
9 |
-
from remfx.utils import FADLoss
|
10 |
-
|
11 |
from umx.openunmix.model import OpenUnmix, Separator
|
12 |
-
|
|
|
13 |
|
14 |
|
15 |
class RemFXModel(pl.LightningModule):
|
@@ -90,9 +94,9 @@ class RemFXModel(pl.LightningModule):
|
|
90 |
return loss
|
91 |
|
92 |
def common_step(self, batch, batch_idx, mode: str = "train"):
|
93 |
-
|
|
|
94 |
self.log(f"{mode}_loss", loss)
|
95 |
-
x, y, label = batch
|
96 |
# Metric logging
|
97 |
with torch.no_grad():
|
98 |
for metric in self.metrics:
|
@@ -119,7 +123,7 @@ class RemFXModel(pl.LightningModule):
|
|
119 |
def on_train_batch_start(self, batch, batch_idx):
|
120 |
# Log initial audio
|
121 |
if self.log_train_audio:
|
122 |
-
x, y,
|
123 |
# Concat samples together for easier viewing in dashboard
|
124 |
input_samples = rearrange(x, "b c t -> c (b t)").unsqueeze(0)
|
125 |
target_samples = rearrange(y, "b c t -> c (b t)").unsqueeze(0)
|
@@ -141,7 +145,7 @@ class RemFXModel(pl.LightningModule):
|
|
141 |
self.log_train_audio = False
|
142 |
|
143 |
def on_validation_batch_start(self, batch, batch_idx, dataloader_idx):
|
144 |
-
x, target,
|
145 |
# Log Input Metrics
|
146 |
for metric in self.metrics:
|
147 |
# SISDR returns negative values, so negate them
|
@@ -185,7 +189,7 @@ class RemFXModel(pl.LightningModule):
|
|
185 |
def on_test_batch_start(self, batch, batch_idx, dataloader_idx):
|
186 |
self.on_validation_batch_start(batch, batch_idx, dataloader_idx)
|
187 |
# Log FAD
|
188 |
-
x, target,
|
189 |
self.log(
|
190 |
"Input_FAD",
|
191 |
self.metrics["FAD"](x, target),
|
@@ -233,7 +237,7 @@ class OpenUnmixModel(torch.nn.Module):
|
|
233 |
self.l1loss = torch.nn.L1Loss()
|
234 |
|
235 |
def forward(self, batch):
|
236 |
-
x, target
|
237 |
X = spectrogram(x, self.window, self.n_fft, self.hop_length, self.alpha)
|
238 |
Y = self.model(X)
|
239 |
sep_out = self.separator(x).squeeze(1)
|
@@ -256,7 +260,7 @@ class DemucsModel(torch.nn.Module):
|
|
256 |
self.l1loss = torch.nn.L1Loss()
|
257 |
|
258 |
def forward(self, batch):
|
259 |
-
x, target
|
260 |
output = self.model(x).squeeze(1)
|
261 |
loss = self.mrstftloss(output, target) + self.l1loss(output, target) * 100
|
262 |
return loss, output
|
@@ -271,7 +275,7 @@ class DiffusionGenerationModel(nn.Module):
|
|
271 |
self.model = DiffusionModel(in_channels=n_channels)
|
272 |
|
273 |
def forward(self, batch):
|
274 |
-
x, target
|
275 |
sampled_out = self.model.sample(x)
|
276 |
return self.model(x), sampled_out
|
277 |
|
@@ -326,3 +330,215 @@ def spectrogram(
|
|
326 |
X = X.view(bs, chs, X.shape[-2], X.shape[-1])
|
327 |
|
328 |
return torch.pow(X.abs() + 1e-8, alpha)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import wandb
|
2 |
import torch
|
3 |
+
import torchaudio
|
4 |
+
import torchmetrics
|
5 |
import pytorch_lightning as pl
|
6 |
+
import torch.nn.functional as F
|
7 |
+
|
8 |
+
from torch import Tensor, nn
|
9 |
from einops import rearrange
|
10 |
+
from torchaudio.models import HDemucs
|
11 |
from audio_diffusion_pytorch import DiffusionModel
|
12 |
from auraloss.time import SISDRLoss
|
13 |
from auraloss.freq import MultiResolutionSTFTLoss
|
|
|
|
|
14 |
from umx.openunmix.model import OpenUnmix, Separator
|
15 |
+
|
16 |
+
from remfx.utils import FADLoss
|
17 |
|
18 |
|
19 |
class RemFXModel(pl.LightningModule):
|
|
|
94 |
return loss
|
95 |
|
96 |
def common_step(self, batch, batch_idx, mode: str = "train"):
|
97 |
+
x, y, _, _ = batch
|
98 |
+
loss, output = self.model((x, y))
|
99 |
self.log(f"{mode}_loss", loss)
|
|
|
100 |
# Metric logging
|
101 |
with torch.no_grad():
|
102 |
for metric in self.metrics:
|
|
|
123 |
def on_train_batch_start(self, batch, batch_idx):
|
124 |
# Log initial audio
|
125 |
if self.log_train_audio:
|
126 |
+
x, y, _, _ = batch
|
127 |
# Concat samples together for easier viewing in dashboard
|
128 |
input_samples = rearrange(x, "b c t -> c (b t)").unsqueeze(0)
|
129 |
target_samples = rearrange(y, "b c t -> c (b t)").unsqueeze(0)
|
|
|
145 |
self.log_train_audio = False
|
146 |
|
147 |
def on_validation_batch_start(self, batch, batch_idx, dataloader_idx):
|
148 |
+
x, target, _, _ = batch
|
149 |
# Log Input Metrics
|
150 |
for metric in self.metrics:
|
151 |
# SISDR returns negative values, so negate them
|
|
|
189 |
def on_test_batch_start(self, batch, batch_idx, dataloader_idx):
|
190 |
self.on_validation_batch_start(batch, batch_idx, dataloader_idx)
|
191 |
# Log FAD
|
192 |
+
x, target, _, _ = batch
|
193 |
self.log(
|
194 |
"Input_FAD",
|
195 |
self.metrics["FAD"](x, target),
|
|
|
237 |
self.l1loss = torch.nn.L1Loss()
|
238 |
|
239 |
def forward(self, batch):
|
240 |
+
x, target = batch
|
241 |
X = spectrogram(x, self.window, self.n_fft, self.hop_length, self.alpha)
|
242 |
Y = self.model(X)
|
243 |
sep_out = self.separator(x).squeeze(1)
|
|
|
260 |
self.l1loss = torch.nn.L1Loss()
|
261 |
|
262 |
def forward(self, batch):
|
263 |
+
x, target = batch
|
264 |
output = self.model(x).squeeze(1)
|
265 |
loss = self.mrstftloss(output, target) + self.l1loss(output, target) * 100
|
266 |
return loss, output
|
|
|
275 |
self.model = DiffusionModel(in_channels=n_channels)
|
276 |
|
277 |
def forward(self, batch):
|
278 |
+
x, target = batch
|
279 |
sampled_out = self.model.sample(x)
|
280 |
return self.model(x), sampled_out
|
281 |
|
|
|
330 |
X = X.view(bs, chs, X.shape[-2], X.shape[-1])
|
331 |
|
332 |
return torch.pow(X.abs() + 1e-8, alpha)
|
333 |
+
|
334 |
+
|
335 |
+
# adapted from https://github.com/qiuqiangkong/audioset_tagging_cnn/blob/master/pytorch/models.py
|
336 |
+
|
337 |
+
|
338 |
+
def init_layer(layer):
|
339 |
+
"""Initialize a Linear or Convolutional layer."""
|
340 |
+
nn.init.xavier_uniform_(layer.weight)
|
341 |
+
|
342 |
+
if hasattr(layer, "bias"):
|
343 |
+
if layer.bias is not None:
|
344 |
+
layer.bias.data.fill_(0.0)
|
345 |
+
|
346 |
+
|
347 |
+
def init_bn(bn):
|
348 |
+
"""Initialize a Batchnorm layer."""
|
349 |
+
bn.bias.data.fill_(0.0)
|
350 |
+
bn.weight.data.fill_(1.0)
|
351 |
+
|
352 |
+
|
353 |
+
class ConvBlock(nn.Module):
|
354 |
+
def __init__(self, in_channels, out_channels):
|
355 |
+
super(ConvBlock, self).__init__()
|
356 |
+
|
357 |
+
self.conv1 = nn.Conv2d(
|
358 |
+
in_channels=in_channels,
|
359 |
+
out_channels=out_channels,
|
360 |
+
kernel_size=(3, 3),
|
361 |
+
stride=(1, 1),
|
362 |
+
padding=(1, 1),
|
363 |
+
bias=False,
|
364 |
+
)
|
365 |
+
|
366 |
+
self.conv2 = nn.Conv2d(
|
367 |
+
in_channels=out_channels,
|
368 |
+
out_channels=out_channels,
|
369 |
+
kernel_size=(3, 3),
|
370 |
+
stride=(1, 1),
|
371 |
+
padding=(1, 1),
|
372 |
+
bias=False,
|
373 |
+
)
|
374 |
+
|
375 |
+
self.bn1 = nn.BatchNorm2d(out_channels)
|
376 |
+
self.bn2 = nn.BatchNorm2d(out_channels)
|
377 |
+
|
378 |
+
self.init_weight()
|
379 |
+
|
380 |
+
def init_weight(self):
|
381 |
+
init_layer(self.conv1)
|
382 |
+
init_layer(self.conv2)
|
383 |
+
init_bn(self.bn1)
|
384 |
+
init_bn(self.bn2)
|
385 |
+
|
386 |
+
def forward(self, input, pool_size=(2, 2), pool_type="avg"):
|
387 |
+
x = input
|
388 |
+
x = F.relu_(self.bn1(self.conv1(x)))
|
389 |
+
x = F.relu_(self.bn2(self.conv2(x)))
|
390 |
+
if pool_type == "max":
|
391 |
+
x = F.max_pool2d(x, kernel_size=pool_size)
|
392 |
+
elif pool_type == "avg":
|
393 |
+
x = F.avg_pool2d(x, kernel_size=pool_size)
|
394 |
+
elif pool_type == "avg+max":
|
395 |
+
x1 = F.avg_pool2d(x, kernel_size=pool_size)
|
396 |
+
x2 = F.max_pool2d(x, kernel_size=pool_size)
|
397 |
+
x = x1 + x2
|
398 |
+
else:
|
399 |
+
raise Exception("Incorrect argument!")
|
400 |
+
|
401 |
+
return x
|
402 |
+
|
403 |
+
|
404 |
+
class Cnn14(nn.Module):
|
405 |
+
def __init__(
|
406 |
+
self,
|
407 |
+
num_classes: int,
|
408 |
+
sample_rate: float,
|
409 |
+
n_fft: int = 2048,
|
410 |
+
hop_length: int = 512,
|
411 |
+
n_mels: int = 128,
|
412 |
+
):
|
413 |
+
super().__init__()
|
414 |
+
self.num_classes = num_classes
|
415 |
+
self.n_fft = n_fft
|
416 |
+
self.hop_length = hop_length
|
417 |
+
|
418 |
+
window = torch.hann_window(n_fft)
|
419 |
+
self.register_buffer("window", window)
|
420 |
+
|
421 |
+
self.melspec = torchaudio.transforms.MelSpectrogram(
|
422 |
+
sample_rate,
|
423 |
+
n_fft,
|
424 |
+
hop_length=hop_length,
|
425 |
+
n_mels=n_mels,
|
426 |
+
)
|
427 |
+
|
428 |
+
self.bn0 = nn.BatchNorm2d(n_mels)
|
429 |
+
|
430 |
+
self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
|
431 |
+
self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
|
432 |
+
self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
|
433 |
+
self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
|
434 |
+
self.conv_block5 = ConvBlock(in_channels=512, out_channels=1024)
|
435 |
+
self.conv_block6 = ConvBlock(in_channels=1024, out_channels=2048)
|
436 |
+
|
437 |
+
self.fc1 = nn.Linear(2048, 2048, bias=True)
|
438 |
+
self.fc_audioset = nn.Linear(2048, num_classes, bias=True)
|
439 |
+
|
440 |
+
self.init_weight()
|
441 |
+
|
442 |
+
def init_weight(self):
|
443 |
+
init_bn(self.bn0)
|
444 |
+
init_layer(self.fc1)
|
445 |
+
init_layer(self.fc_audioset)
|
446 |
+
|
447 |
+
def forward(self, x: torch.Tensor):
|
448 |
+
"""
|
449 |
+
Input: (batch_size, data_length)"""
|
450 |
+
|
451 |
+
x = self.melspec(x)
|
452 |
+
x = x.permute(0, 2, 1, 3)
|
453 |
+
x = self.bn0(x)
|
454 |
+
x = x.permute(0, 2, 1, 3)
|
455 |
+
|
456 |
+
if self.training:
|
457 |
+
pass
|
458 |
+
# x = self.spec_augmenter(x)
|
459 |
+
|
460 |
+
x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
|
461 |
+
x = F.dropout(x, p=0.2, training=self.training)
|
462 |
+
x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
|
463 |
+
x = F.dropout(x, p=0.2, training=self.training)
|
464 |
+
x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
|
465 |
+
x = F.dropout(x, p=0.2, training=self.training)
|
466 |
+
x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
|
467 |
+
x = F.dropout(x, p=0.2, training=self.training)
|
468 |
+
x = self.conv_block5(x, pool_size=(2, 2), pool_type="avg")
|
469 |
+
x = F.dropout(x, p=0.2, training=self.training)
|
470 |
+
x = self.conv_block6(x, pool_size=(1, 1), pool_type="avg")
|
471 |
+
x = F.dropout(x, p=0.2, training=self.training)
|
472 |
+
x = torch.mean(x, dim=3)
|
473 |
+
|
474 |
+
(x1, _) = torch.max(x, dim=2)
|
475 |
+
x2 = torch.mean(x, dim=2)
|
476 |
+
x = x1 + x2
|
477 |
+
x = F.dropout(x, p=0.5, training=self.training)
|
478 |
+
x = F.relu_(self.fc1(x))
|
479 |
+
clipwise_output = self.fc_audioset(x)
|
480 |
+
|
481 |
+
return clipwise_output
|
482 |
+
|
483 |
+
|
484 |
+
class FXClassifier(pl.LightningModule):
|
485 |
+
def __init__(
|
486 |
+
self,
|
487 |
+
lr: float,
|
488 |
+
lr_weight_decay: float,
|
489 |
+
sample_rate: float,
|
490 |
+
network: nn.Module,
|
491 |
+
):
|
492 |
+
super().__init__()
|
493 |
+
self.lr = lr
|
494 |
+
self.lr_weight_decay = lr_weight_decay
|
495 |
+
self.sample_rate = sample_rate
|
496 |
+
self.network = network
|
497 |
+
|
498 |
+
def forward(self, x: torch.Tensor):
|
499 |
+
return self.network(x)
|
500 |
+
|
501 |
+
def common_step(self, batch, batch_idx, mode: str = "train"):
|
502 |
+
x, y, dry_label, wet_label = batch
|
503 |
+
pred_label = self.network(x)
|
504 |
+
loss = torch.nn.functional.cross_entropy(pred_label, dry_label)
|
505 |
+
self.log(
|
506 |
+
f"{mode}_loss",
|
507 |
+
loss,
|
508 |
+
on_step=True,
|
509 |
+
on_epoch=True,
|
510 |
+
prog_bar=True,
|
511 |
+
logger=True,
|
512 |
+
sync_dist=True,
|
513 |
+
)
|
514 |
+
|
515 |
+
self.log(
|
516 |
+
f"{mode}_mAP",
|
517 |
+
torchmetrics.functional.retrieval_average_precision(
|
518 |
+
pred_label, dry_label.long()
|
519 |
+
),
|
520 |
+
on_step=True,
|
521 |
+
on_epoch=True,
|
522 |
+
prog_bar=True,
|
523 |
+
logger=True,
|
524 |
+
sync_dist=True,
|
525 |
+
)
|
526 |
+
|
527 |
+
return loss
|
528 |
+
|
529 |
+
def training_step(self, batch, batch_idx):
|
530 |
+
return self.common_step(batch, batch_idx, mode="train")
|
531 |
+
|
532 |
+
def validation_step(self, batch, batch_idx):
|
533 |
+
return self.common_step(batch, batch_idx, mode="valid")
|
534 |
+
|
535 |
+
def test_step(self, batch, batch_idx):
|
536 |
+
return self.common_step(batch, batch_idx, mode="test")
|
537 |
+
|
538 |
+
def configure_optimizers(self):
|
539 |
+
optimizer = torch.optim.AdamW(
|
540 |
+
self.network.parameters(),
|
541 |
+
lr=self.lr,
|
542 |
+
weight_decay=self.lr_weight_decay,
|
543 |
+
)
|
544 |
+
return optimizer
|