forked from CPJKU/dcase2024_task1_baseline
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_script.py
227 lines (203 loc) · 9.61 KB
/
test_script.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
#### This is the script to generate h5 files provided by Joseph####
import pytorch_lightning as pl
from pytorch_lightning.loggers import WandbLogger
# import torch
import torchaudio
from torch.utils.data import DataLoader, BatchSampler
# import argparse
import torch.nn.functional as F
# import transformers
# import wandb
# import json
# from dataset.dcase24 import get_training_set, get_test_set, get_eval_set
# from dataset.ntu_dcase24_v1 import ntu_get_training_set, ntu_get_training_set_dir, ntu_get_test_set, ntu_gen_base_training_h5, open_h5, close_h5
# from helpers.init import worker_init_fn
# from models.baseline import get_model
# # from models.ntu_baseline import get_ntu_model, get_ntu_protonet
# from helpers.utils import mixstyle, ntu_mixstyle
# from helpers import nessi
# import shutil
# import pathlib
import os
# from helpers.ntu_data_sampler import ProtoNetDataSampler
# from dataset.ntu_dcase24 import SimpleSelectionDataset, BasicDCASE24DatasetProtoNet
import pandas as pd
from argparse import Namespace
import h5py
from tqdm import tqdm
from models.mel import AugmentMelSTFT
# import glob
args = Namespace(project_name='DCASE24_BCBL',
experiment_name='Protonet-NTU',
num_workers=0, precision='32',
evaluate=False, ckpt_id='057o1jd1',
orig_sample_rate=44100,
subset=5, n_classes=10,
in_channels=1,
base_channels=20,
channels_multiplier=1.5,
expansion_rate=2,
n_epochs=1,
batch_size=256,
mixstyle_p=0.4,
mixstyle_alpha=0.3,
weight_decay=0.00001,
roll_sec=0, lr=0.01,
dir_p = 0.6,
warmup_steps=20,
sample_rate=44100,
resample_rate = 44100,
window_length=3072,
hop_length=500,
n_fft=4096,
n_mels=256,
freqm=0,
timem=0,
f_min=0,
f_max=None,
fmin_aug_range=1,
fmax_aug_range=1000,
way=10,
shot=5,
query=10,
episode=1000)
# args = Namespace(project_name='DCASE24_Task1_PaSST',
# experiment_name='PaSST-NTU',
# num_workers=0, precision='32',
# evaluate=False, ckpt_id='057o1jd1',
# orig_sample_rate=44100,
# subset=5, n_classes=13,
# in_channels=1,
# base_channels=20,
# channels_multiplier=1.5,
# expansion_rate=2,
# n_epochs=1,
# batch_size=256,
# mixstyle_p=0.4,
# mixstyle_alpha=0.3,
# weight_decay=0.00001,
# roll_sec=0, lr=0.01,
# dir_p = 0.6,
# warmup_steps=20,##################### Start Here #####################
# resample_rate = 44100,
# sample_rate=44100,
# window_length=800,
# hop_length=320,
# n_fft=1024,
# n_mels=128,
# freqm=0,
# timem=0, ##################### Done ########################
# f_min=0,
# f_max=None,
# fmin_aug_range=1,
# fmax_aug_range=1000,
# way=10,
# shot=5,
# query=10,
# episode=1000)
config = args
# meta_csv = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2022-mobile-development\meta.csv" # DSP
meta_csv = r"D:\Sean\DCASE\datasets\Extract_to_Folder\TAU-urban-acoustic-scenes-2022-mobile-development\meta.csv" # ALI
# meta_csv = r"F:\CochlScene\meta.csv" # DSP
# meta_csv = r"D:\Sean\CochlScene\meta.csv" # ALI
# train_files_csv = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2022-mobile-development\split100.csv" # DSP
train_files_csv = r"D:\Sean\DCASE\datasets\Extract_to_Folder\TAU-urban-acoustic-scenes-2022-mobile-development\split100.csv" # ALI
# train_files_csv = r"F:\Github\dcase2024_task1\split_setup\splitcochl.csv"
# train_files_csv = r"D:\Sean\github\cpjku_dcase23_NTU\split_setup\splitcochl10s.csv"
# eval_meta_csv = 'c:/Dataset/eval_dataset_2024/meta.csv'
# dataset_dir = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2022-mobile-development" # DSP
dataset_dir = r"D:\Sean\DCASE\datasets\Extract_to_Folder\TAU-urban-acoustic-scenes-2022-mobile-development" # ALI
# dataset_dir = r"F:\CochlScene"
# dataset_dir = r"D:\Sean\CochlScene" # ALI
# eval_dataset_dir = 'c:/Dataset/eva_dataset_2024/'
# eval_csv = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2024-mobile-evaluation\evaluation_setup\fold1_test.csv" # DSP
eval_csv = r"D:\Sean\DCASE\datasets\Extract_to_Folder\TAU-urban-acoustic-scenes-2022-mobile-development\evaluation_setup\fold1_test.csv" #ALI
# eval_csv = r"F:\Github\dcase2024_task1\split_setup\val_cochl10s.csv" #DSP
# eval_csv = r"D:\Sean\github\cpjku_dcase23_NTU\split_setup\val_cochl10s.csv" # ALI
# eval_dir = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2024-mobile-evaluation" # DSP
# eval_dir = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2024-mobile-evaluation" # AlI
# test_csv = r"F:\Github\dcase2024_task1\split_setup\test_cochl10s.csv" # DSP
test_csv = r"D:\Sean\github\dcase2024_task1\split_setup\test.csv" # ALI
# test_csv = r"D:\Sean\github\cpjku_dcase23_NTU\split_setup\test_cochl10s.csv" # ALI
# eval_dir = r"F:\DCASE\2024\Datasets\TAU-urban-acoustic-scenes-2024-mobile-evaluation"
dataset_config = {
"dataset_name": "tau24",
"meta_csv": os.path.join(dataset_dir, "meta.csv"),
"split_path": "split_setup",
"split_url": "https://github.com/CPJKU/dcase2024_task1_baseline/releases/download/files/",
"test_split_csv": "test.csv",
"eval_dir": os.path.join(dataset_dir, "..", "eval_dataset_2024"),
"eval_meta_csv": os.path.join(dataset_dir, "..", "eval_dataset_2024", "meta.csv"),
"dirs_path": 'C:/Dataset/mic_Impulse/',
}
mel = torchaudio.transforms.MelSpectrogram(
sample_rate=config.sample_rate,
n_fft=config.n_fft,
win_length=config.window_length,
hop_length=config.hop_length,
n_mels=config.n_mels,
f_min=config.f_min,
f_max=config.f_max
)
mel_passt = AugmentMelSTFT(n_mels=config.n_mels,
sr=config.resample_rate,
win_length=config.window_length,
hopsize=config.hop_length,
n_fft=config.n_fft,
freqm=config.freqm,
timem=config.timem,
fmin=config.f_min,
fmax=config.f_max,
fmin_aug_range=config.fmin_aug_range,
fmax_aug_range=config.fmax_aug_range
)
# # to create audio samples h5 file
df = pd.read_csv(meta_csv, sep="\t")
train_files = pd.read_csv(train_files_csv, sep='\t')['filename'].values.reshape(-1)
files = df['filename'].values.reshape(-1)
hf = h5py.File('h5py_cochl_wav', 'w')
for file_idx in tqdm(range(len(files))):
mel_sig, _ = torchaudio.load(os.path.join(dataset_dir, files[file_idx]))
#output_str = dataset_dir + 'h5' + train_files[file_idx][5:-4] + '.h5'
output_str = files[file_idx][5:-4]
print(f"output = {output_str}")
#with h5py.File(output_str, 'w') as hf:
hf.create_dataset(output_str, data = mel_sig)
hf.close()
# Create mel HDF5 file
# df = pd.read_csv(meta_csv, sep="\t")
# train_files = pd.read_csv(train_files_csv, sep='\t')['filename'].values.reshape(-1)
# files = df['filename'].values.reshape(-1)
# hf = h5py.File('h5py_cochl10_256bins', 'w')
# for file_idx in tqdm(range(len(files))):
# sig, _ = torchaudio.load(os.path.join(dataset_dir, files[file_idx]))
# mel_sig = mel_passt(sig)
# output_str = files[file_idx][5:-4]
# hf.create_dataset(output_str, data=mel_sig)
# hf.close()
# # to create mel data h5 file
# df = pd.read_csv(meta_csv, sep="\t")
# #train_files = pd.read_csv(train_files_csv, sep='\t')['filename'].values.reshape(-1)
# train_files = pd.read_csv(train_files_csv, sep='\t')['filename'].values.reshape(-1)
# files = df[['filename']].values.reshape(-1)
# hf = h5py.File('h5py_mel_256bins', 'w')
# for file_idx in tqdm(range(len(files))):
# sig, _ = torchaudio.load(os.path.join(dataset_dir, files[file_idx]))
# mel_sig = mel(sig)
# #output_str = dataset_dir + 'h5' + train_files[file_idx][5:-4] + '.h5'
# output_str = files[file_idx][5:-4]
# #with h5py.File(output_str, 'w') as hf:
# hf.create_dataset(output_str, data = mel_sig)
# hf.close()
# # Save eval set files to h5
# df = pd.read_csv(eval_csv, sep="\t")
# # train_files = pd.read_csv(train_files_csv, sep='\t')['filename'].values.reshape(-1)
# files = df['filename'].values.reshape(-1)
# hf = h5py.File('h5py_audio_eval_wav', 'w')
# for file_idx in tqdm(range(len(files))):
# mel_sig, _ = torchaudio.load(os.path.join(eval_dir, files[file_idx]))
# #output_str = dataset_dir + 'h5' + train_files[file_idx][5:-4] + '.h5'
# output_str = files[file_idx][5:-4]
# #with h5py.File(output_str, 'w') as hf:
# hf.create_dataset(output_str, data = mel_sig)
# hf.close()