You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
PaddleSpeech/examples/other/1xt2x/baidu_en8k/conf/deepspeech2.yaml

65 lines
1.5 KiB

# https://yaml.org/type/float.html
###########################################
# Data #
###########################################
train_manifest: data/manifest.train
dev_manifest: data/manifest.dev
test_manifest: data/manifest.test-clean
min_input_len: 0.0
max_input_len: .inf # second
min_output_len: 0.0
max_output_len: .inf
min_output_input_ratio: 0.00
max_output_input_ratio: .inf
###########################################
# Dataloader #
###########################################
batch_size: 64 # one gpu
mean_std_filepath: data/mean_std.npz
unit_type: char
vocab_filepath: data/vocab.txt
augmentation_config: conf/augmentation.json
random_seed: 0
spm_model_prefix:
spectrum_type: linear
feat_dim:
delta_delta: False
stride_ms: 10.0
window_ms: 20.0
n_fft: None
max_freq: None
target_sample_rate: 16000
use_dB_normalization: True
target_dB: -20
dither: 1.0
keep_transcription_text: False
sortagrad: True
shuffle_method: batch_shuffle
num_workers: 2
############################################
# Network Architecture #
############################################
num_conv_layers: 2
num_rnn_layers: 3
rnn_layer_size: 1024
use_gru: True
share_rnn_weights: False
blank_id: 28
###########################################
# Training #
###########################################
n_epoch: 80
accum_grad: 1
lr: 2e-3
lr_decay: 0.83
weight_decay: 1e-06
global_grad_clip: 3.0
log_interval: 100
checkpoint:
kbest_n: 50
latest_n: 5