data_config = {
"train": {
"x_path": "drive/MyDrive/x_train.wav",
"y_path": "drive/MyDrive/y_train.wav",
"ny": 8192
},
"validation": {
"x_path": "drive/MyDrive/x_test.wav",
"y_path": "drive/MyDrive/y_test.wav",
"ny": None
},
"common": {
"delay": int(input("What is the latency (in samples) of your reamp? "))
}
}
model_config = { # rickard@tutanota.com
"net": {
"name": "WaveNet",
"config": {
"layers_configs": [
{
"input_size": 1,
"condition_size": 1,
"channels": 8,
"head_size": 8,
"kernel_size": 5,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 8,
"channels": 8,
"head_size": 8,
"kernel_size": 5,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 8,
"channels": 8,
"head_size": 8,
"kernel_size": 5,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 8,
"channels": 8,
"head_size": 8,
"kernel_size": 5,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 8,
"channels": 8,
"head_size": 1,
"kernel_size": 5,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": True
}
],
"head_scale": 0.99
}
},
"loss": {
"val_loss": "esr",
"pre_emph_mrstft_weight": 0.0002,
"pre_emph_mrstft_coef": 0.85
},
"optimizer": {
"lr": 0.002,
},
"lr_scheduler": {
"class": "ExponentialLR",
"seed": 42,
"kwargs": {"gamma": 0.9985},
}
}
learning_config = {
"train_dataloader": {
"batch_size": 16,
"shuffle": True,
"pin_memory": True,
"drop_last": True,
"num_workers": 8,
},
"val_dataloader": {},
"trainer": {
"accelerator": "gpu",
"devices": 1,
"max_epochs": 1500
}
}
|
{
"head_scale": 0.99,
"layers_configs": [
{
"gated": false,
"channels": 8,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 8,
"activation": "Tanh",
"input_size": 1,
"kernel_size": 5,
"condition_size": 1
},
{
"gated": false,
"channels": 8,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 8,
"activation": "Tanh",
"input_size": 8,
"kernel_size": 5,
"condition_size": 1
},
{
"gated": false,
"channels": 8,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 8,
"activation": "Tanh",
"input_size": 8,
"kernel_size": 5,
"condition_size": 1
},
{
"gated": false,
"channels": 8,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 8,
"activation": "Tanh",
"input_size": 8,
"kernel_size": 5,
"condition_size": 1
},
{
"gated": false,
"channels": 8,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": true,
"head_size": 1,
"activation": "Tanh",
"input_size": 8,
"kernel_size": 5,
"condition_size": 1
}
]
}
|
data_config = {
"train": {
"x_path": "drive/MyDrive/x_train.wav",
"y_path": "drive/MyDrive/y_train.wav",
"ny": 8192
},
"validation": {
"x_path": "drive/MyDrive/x_test.wav",
"y_path": "drive/MyDrive/y_test.wav",
"ny": None
},
"common": {
"delay": int(input("What is the latency (in samples) of your reamp? "))
}
}
model_config = { # rickard@tutanota.com
"net": {
"name": "WaveNet",
"config": {
"layers_configs": [
{
"input_size": 1,
"condition_size": 1,
"channels": 10,
"head_size": 10,
"kernel_size": 6,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 10,
"channels": 10,
"head_size": 10,
"kernel_size": 6,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 10,
"channels": 10,
"head_size": 10,
"kernel_size": 6,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 10,
"channels": 10,
"head_size": 10,
"kernel_size": 6,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": False
},
{
"condition_size": 1,
"input_size": 10,
"channels": 10,
"head_size": 1,
"kernel_size": 6,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"activation": "Tanh",
"gated": False,
"head_bias": True
}
],
"head_scale": 0.99
}
},
"loss": {
"val_loss": "esr",
"pre_emph_mrstft_weight": 0.0002,
"pre_emph_mrstft_coef": 0.85
},
"optimizer": {
"lr": 0.002,
},
"lr_scheduler": {
"class": "ExponentialLR",
"seed": 42,
"kwargs": {"gamma": 0.9985},
}
}
learning_config = {
"train_dataloader": {
"batch_size": 16,
"shuffle": True,
"pin_memory": True,
"drop_last": True,
"num_workers": 8,
},
"val_dataloader": {},
"trainer": {
"accelerator": "gpu",
"devices": 1,
"max_epochs": 1500
}
}
|
{
"head_scale": 0.99,
"layers_configs": [
{
"gated": false,
"channels": 10,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 10,
"activation": "Tanh",
"input_size": 1,
"kernel_size": 6,
"condition_size": 1
},
{
"gated": false,
"channels": 10,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 10,
"activation": "Tanh",
"input_size": 10,
"kernel_size": 6,
"condition_size": 1
},
{
"gated": false,
"channels": 10,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 10,
"activation": "Tanh",
"input_size": 10,
"kernel_size": 6,
"condition_size": 1
},
{
"gated": false,
"channels": 10,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": false,
"head_size": 10,
"activation": "Tanh",
"input_size": 10,
"kernel_size": 6,
"condition_size": 1
},
{
"gated": false,
"channels": 10,
"dilations": [
1024,
256,
64,
16,
4,
1
],
"head_bias": true,
"head_size": 1,
"activation": "Tanh",
"input_size": 10,
"kernel_size": 6,
"condition_size": 1
}
]
}
|