|
|
|
|
# https://yaml.org/type/float.html
|
|
|
|
|
###########################################
|
|
|
|
|
# Data #
|
|
|
|
|
###########################################
|
|
|
|
|
dataset: 'paddlespeech.audio.datasets:HeySnips'
|
|
|
|
|
data_dir: '/PATH/TO/DATA/hey_snips_research_6k_en_train_eval_clean_ter'
|
|
|
|
|
|
|
|
|
|
############################################
|
|
|
|
|
# Network Architecture #
|
|
|
|
|
############################################
|
|
|
|
|
backbone: 'paddlespeech.kws.models:MDTC'
|
|
|
|
|
num_keywords: 1
|
|
|
|
|
stack_num: 3
|
|
|
|
|
stack_size: 4
|
|
|
|
|
in_channels: 80
|
|
|
|
|
res_channels: 32
|
|
|
|
|
kernel_size: 5
|
|
|
|
|
|
|
|
|
|
###########################################
|
|
|
|
|
# Feature #
|
|
|
|
|
###########################################
|
|
|
|
|
feat_type: 'kaldi_fbank'
|
|
|
|
|
sample_rate: 16000
|
|
|
|
|
frame_shift: 10
|
|
|
|
|
frame_length: 25
|
|
|
|
|
n_mels: 80
|
|
|
|
|
|
|
|
|
|
###########################################
|
|
|
|
|
# Training #
|
|
|
|
|
###########################################
|
|
|
|
|
epochs: 100
|
|
|
|
|
num_workers: 16
|
|
|
|
|
batch_size: 100
|
|
|
|
|
checkpoint_dir: './checkpoint'
|
|
|
|
|
save_freq: 10
|
|
|
|
|
log_freq: 10
|
|
|
|
|
learning_rate: 0.001
|
|
|
|
|
weight_decay: 0.00005
|
|
|
|
|
grad_clip: 5.0
|
|
|
|
|
|
|
|
|
|
###########################################
|
|
|
|
|
# Scoring #
|
|
|
|
|
###########################################
|
|
|
|
|
batch_size: 100
|
|
|
|
|
num_workers: 16
|
|
|
|
|
checkpoint: './checkpoint/epoch_100/model.pdparams'
|
|
|
|
|
score_file: './scores.txt'
|
|
|
|
|
stats_file: './stats.0.txt'
|
|
|
|
|
img_file: './det.png'
|
