unlinken of already processed training files (seed dependent audio augmentation)

_paramters.py

@@ -31,12 +31,12 @@ main_arg_parser.add_argument("--data_mixup", type=strtobool, default=False, help
 main_arg_parser.add_argument("--data_stretch", type=strtobool, default=True, help="")
 
 # Transformation Parameters
-main_arg_parser.add_argument("--data_loudness_ratio", type=float, default=0.4, help="")  # 0.4
+main_arg_parser.add_argument("--data_loudness_ratio", type=float, default=0, help="")  # 0.4
-main_arg_parser.add_argument("--data_shift_ratio", type=float, default=0.3, help="")  # 0.3
+main_arg_parser.add_argument("--data_shift_ratio", type=float, default=0, help="")  # 0.3
-main_arg_parser.add_argument("--data_noise_ratio", type=float, default=0.4, help="")  # 0.4
+main_arg_parser.add_argument("--data_noise_ratio", type=float, default=0, help="")  # 0.4
-main_arg_parser.add_argument("--data_mask_ratio", type=float, default=0.2, help="")  # 0.2
+main_arg_parser.add_argument("--data_mask_ratio", type=float, default=0, help="")  # 0.2
-main_arg_parser.add_argument("--data_speed_ratio", type=float, default=0.3, help="")  # 0.3
+main_arg_parser.add_argument("--data_speed_ratio", type=float, default=0, help="")  # 0.3
-main_arg_parser.add_argument("--data_speed_factor", type=float, default=0.7, help="")  # 0.7
+main_arg_parser.add_argument("--data_speed_factor", type=float, default=0, help="")  # 0.7
 
 # Training Parameters
 main_arg_parser.add_argument("--train_outpath", type=str, default="output", help="")
@@ -50,8 +50,8 @@ main_arg_parser.add_argument("--train_lr", type=float, default=1e-4, help="")
 main_arg_parser.add_argument("--train_num_sanity_val_steps", type=int, default=0, help="")
 
 # Model Parameters
-main_arg_parser.add_argument("--model_type", type=str, default="BCMC", help="")
+main_arg_parser.add_argument("--model_type", type=str, default="CC", help="")
-main_arg_parser.add_argument("--model_secondary_type", type=str, default="BCMC", help="")
+main_arg_parser.add_argument("--model_secondary_type", type=str, default="CC", help="")
 main_arg_parser.add_argument("--model_weight_init", type=str, default="xavier_normal_", help="")
 main_arg_parser.add_argument("--model_activation", type=str, default="leaky_relu", help="")
 main_arg_parser.add_argument("--model_filters", type=str, default="[32, 64, 128, 64]", help="")

main_inference.py

@@ -7,6 +7,7 @@ import variables as V
 from torch.utils.data import DataLoader, Dataset
 from torchvision.transforms import Compose, RandomApply
 
+from ml_lib.audio_toolset.audio_augmentation import Speed
 from ml_lib.audio_toolset.audio_io import AudioToMel, NormalizeLocal, MelToImage
 
 # Dataset and Dataloaders
@@ -17,6 +18,7 @@ from ml_lib.audio_toolset.mel_augmentation import NoiseInjection, LoudnessManipu
 from ml_lib.utils.logging import Logger
 from ml_lib.utils.model_io import SavedLightningModels
 from ml_lib.utils.transforms import ToTensor
+from ml_lib.visualization.tools import Plotter
 from util.config import MConfig
 
 # Datasets
@@ -25,23 +27,22 @@ from datasets.binar_masks import BinaryMasksDataset
 
 def prepare_dataloader(config_obj):
     mel_transforms = Compose([
-        # Audio to Mel Transformations
+        Speed(0, 0),
         AudioToMel(sr=config_obj.data.sr, n_mels=config_obj.data.n_mels, n_fft=config_obj.data.n_fft,
-                   hop_length=config_obj.data.hop_length), MelToImage()])
+                   hop_length=config_obj.data.hop_length),
+        MelToImage()])
     transforms = Compose([NormalizeLocal(), ToTensor()])
     aug_transforms = Compose([
-        RandomApply([
+            NoiseInjection(0.4),
-            NoiseInjection(config_obj.data.noise_ratio),
+            LoudnessManipulator(0),
-            LoudnessManipulator(config_obj.data.loudness_ratio),
+            ShiftTime(0),
-            ShiftTime(config_obj.data.shift_ratio),
+            MaskAug(0),
-            MaskAug(config_obj.data.mask_ratio),
-        ], p=0.6),
         # Utility
         NormalizeLocal(), ToTensor()
     ])
 
     dataset: Dataset = BinaryMasksDataset(config_obj.data.root, setting='train',
-                                          mel_transforms=mel_transforms, transforms=transforms
+                                          mel_transforms=mel_transforms, transforms=aug_transforms
                                           )
     # noinspection PyTypeChecker
     return DataLoader(dataset, batch_size=None, num_workers=0, shuffle=False)
@@ -69,6 +70,12 @@ if __name__ == '__main__':
     config = MConfig()
     config.read_file((outpath / model_type / parameters / version / config_filename).open('r'))
     test_dataloader = prepare_dataloader(config)
+    p = Plotter(outpath)
+    from matplotlib import pyplot as plt
+
+    d = test_dataloader.dataset[0][0].squeeze()
+    plt.imshow(d)
+
     loaded_model = restore_logger_and_model(config)
     loaded_model.eval()
 

util/metric_reader.py

@@ -12,6 +12,8 @@ config_file_name = 'config.ini'
 
 if __name__ == '__main__':
     for model_path in outpath.iterdir():
+        if not model_path.is_dir():
+            continue
         out_file = (model_path / metric_file_name)
         for paramter_configuration in model_path.iterdir():
             uar_scores = defaultdict(list)
@@ -46,5 +48,8 @@ if __name__ == '__main__':
                 writer = csv.DictWriter(f, delimiter=',', lineterminator='\n', fieldnames=headers)
                 if not file_existed:
                     writer.writeheader()  # file doesn't exist yet, write a header
-                for row_idx in range(len(uar_scores['mean'])):
+                try:
-                    writer.writerow({key: uar_scores[key][row_idx] for key in headers})
+                    for row_idx in range(len(uar_scores['mean'])):
+                        writer.writerow({key: uar_scores[key][row_idx] for key in headers})
+                except IndexError:
+                    print('could not read: ')