File size: 4,911 Bytes
e915e01
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c429442
 
 
 
 
 
 
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
# Generated 2021-12-01 from:
# /home/mila/s/subakany/speechbrain-soundskrit/recipes/WHAMandWHAMR/enhancement/hparams/sepformer-wham.yaml
# yamllint disable
# ################################
# Model: SepFormer for source separation
# https://arxiv.org/abs/2010.13154
#
# Dataset : WHAM!
# ################################
# Basic parameters
# Seed needs to be set at top of yaml, before objects with parameters are made
#
seed: 1234
__set_seed: !apply:torch.manual_seed [1234]

# Data params

# the data folder for the wham dataset
# needs to end with wham_original for the wham dataset
# respecting this convention effects the code functionality
data_folder: /network/tmp1/subakany/wham_original
task: enhancement
dereverberate: false

# the path for wsj0/si_tr_s/ folder -- only needed if dynamic mixing is used
# e.g. /yourpath/wsj0-processed/si_tr_s/
## you need to convert the original wsj0 to 8k
# you can do this conversion with the script ../meta/preprocess_dynamic_mixing.py
base_folder_dm: /network/tmp1/subakany/wsj0-processed/si_tr_s/

experiment_name: sepformer-wham-enhancement
output_folder: results/sepformer-wham-enhancement/1234
train_log: results/sepformer-wham-enhancement/1234/train_log.txt
save_folder: results/sepformer-wham-enhancement/1234/save

# the file names should start with whamr instead of whamorg
train_data: results/sepformer-wham-enhancement/1234/save/whamorg_tr.csv
valid_data: results/sepformer-wham-enhancement/1234/save/whamorg_cv.csv
test_data: results/sepformer-wham-enhancement/1234/save/whamorg_tt.csv
skip_prep: false


# Experiment params
auto_mix_prec: true # Set it to True for mixed precision
test_only: true
num_spks: 1 # set to 3 for wsj0-3mix
progressbar: true
save_audio: false # Save estimated sources on disk
sample_rate: 8000

# Training parameters
N_epochs: 200
batch_size: 1
lr: 0.00015
clip_grad_norm: 5
loss_upper_lim: 999999  # this is the upper limit for an acceptable loss
# if True, the training sequences are cut to a specified length
limit_training_signal_len: false
# this is the length of sequences if we choose to limit
# the signal length of training sequences
training_signal_len: 32000000

# Set it to True to dynamically create mixtures at training time
dynamic_mixing: true

# Parameters for data augmentation
use_wavedrop: false
use_speedperturb: true
use_speedperturb_sameforeachsource: false
use_rand_shift: false
min_shift: -8000
max_shift: 8000

speedperturb: !new:speechbrain.lobes.augment.TimeDomainSpecAugment
  perturb_prob: 1.0
  drop_freq_prob: 0.0
  drop_chunk_prob: 0.0
  sample_rate: 8000
  speeds: [95, 100, 105]

wavedrop: !new:speechbrain.lobes.augment.TimeDomainSpecAugment
  perturb_prob: 0.0
  drop_freq_prob: 1.0
  drop_chunk_prob: 1.0
  sample_rate: 8000

# loss thresholding -- this thresholds the training loss
threshold_byloss: true
threshold: -30

# Encoder parameters
N_encoder_out: 256
out_channels: 256
kernel_size: 16
kernel_stride: 8

# Dataloader options
dataloader_opts:
  batch_size: 1
  num_workers: 3


# Specifying the network
Encoder: &id003 !new:speechbrain.lobes.models.dual_path.Encoder
  kernel_size: 16
  out_channels: 256


SBtfintra: &id001 !new:speechbrain.lobes.models.dual_path.SBTransformerBlock
  num_layers: 8
  d_model: 256
  nhead: 8
  d_ffn: 1024
  dropout: 0
  use_positional_encoding: true
  norm_before: true

SBtfinter: &id002 !new:speechbrain.lobes.models.dual_path.SBTransformerBlock
  num_layers: 8
  d_model: 256
  nhead: 8
  d_ffn: 1024
  dropout: 0
  use_positional_encoding: true
  norm_before: true

MaskNet: &id005 !new:speechbrain.lobes.models.dual_path.Dual_Path_Model

  num_spks: 1
  in_channels: 256
  out_channels: 256
  num_layers: 2
  K: 250
  intra_model: *id001
  inter_model: *id002
  norm: ln
  linear_layer_after_inter_intra: false
  skip_around_intra: true

Decoder: &id004 !new:speechbrain.lobes.models.dual_path.Decoder
  in_channels: 256
  out_channels: 1
  kernel_size: 16
  stride: 8
  bias: false

optimizer: !name:torch.optim.Adam
  lr: 0.00015
  weight_decay: 0

loss: !name:speechbrain.nnet.losses.get_si_snr_with_pitwrapper

lr_scheduler: &id007 !new:speechbrain.nnet.schedulers.ReduceLROnPlateau

  factor: 0.5
  patience: 2
  dont_halve_until_epoch: 65

epoch_counter: &id006 !new:speechbrain.utils.epoch_loop.EpochCounter
  limit: 200

modules:
  encoder: *id003
  decoder: *id004
  masknet: *id005
checkpointer: !new:speechbrain.utils.checkpoints.Checkpointer
  checkpoints_dir: results/sepformer-wham-enhancement/1234/save
  recoverables:
    encoder: *id003
    decoder: *id004
    masknet: *id005
    counter: *id006
    lr_scheduler: *id007
train_logger: !new:speechbrain.utils.train_logger.FileTrainLogger
  save_file: results/sepformer-wham-enhancement/1234/train_log.txt

pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
  loadables:
    encoder: !ref <Encoder>
    masknet: !ref <MaskNet>
    decoder: !ref <Decoder>