asr_hyperparams.yaml

# Generated 2022-10-03 from:
# /netscratch/sagar/thesis/speechbrain/recipes/CommonVoice_de/ASR-Libri/seq2seq/hparams/train.yaml
# yamllint disable
# ############################################################################
# Model: E2E ASR with attention-based ASR
# Encoder: CRDNN model
# Decoder: GRU + beamsearch + RNNLM
# Tokens: BPE with unigram
# losses: CTC+ NLL
# Training: Librispeech 960h
# Authors:  Ju-Chieh Chou, Mirco Ravanelli, Abdel Heba, Peter Plantinga,
# Samuele Cornell 2020
# ############################################################################

# Seed needs to be set at top of yaml, before objects with parameters
seed: 1200
__set_seed: !apply:torch.manual_seed [1200]
output_folder: results/CRDNN_BPE_960h_LM/1200
wer_file: results/CRDNN_BPE_960h_LM/1200/wer.txt
save_folder: results/CRDNN_BPE_960h_LM/1200/save
train_log: results/CRDNN_BPE_960h_LM/1200/train_log.txt

# Language model (LM) pretraining
# NB: To avoid mismatch, the speech recognizer must be trained with the same
# tokenizer used for LM training. Here, we download everything from the
# speechbrain HuggingFace repository. However, a local path pointing to a
# directory containing the lm.ckpt and tokenizer.ckpt may also be specified
# instead. E.g if you want to use your own LM / tokenizer.
# We have bos/eos id 0/0 so we use the same tokenizer and LM that uses bos id and eos id as 0/0.
pretrained_tokenizer_path: ../../Tokenizer/results/unigram/
pretrained_lm_path: ../../LM/results/RNN/2995/save/CKPT+2022-08-18+18-22-18+00

# Data files
data_folder: ../../CommonVoice      # !PLACEHOLDER
  # e,g./path/to/LibriSpeech
  # noise/ris dataset will automatically be downloaded

# Data files
train_tsv_file: ../../CommonVoice/train.tsv   # Standard CommonVoice .tsv files
dev_tsv_file: ../../CommonVoice/dev.tsv   # Standard CommonVoice .tsv files
test_tsv_file: ../../CommonVoice/test.tsv   # Standard CommonVoice .tsv files
accented_letters: true
language: de
ckpt_interval_minutes: 15 # save checkpoint every N min
csv_dir: ../../cv_de_acc
data_folder_rirs: ../../cv_de_acc # where to store noisy data for augment (change it if needed)
train_csv: ../../cv_de_acc/train.csv
valid_csv: ../../cv_de_acc/dev.csv
test_csv: ../../cv_de_acc/test.csv
skip_prep: false

# Training parameters
number_of_epochs: 25
number_of_ctc_epochs: 5
batch_size: 8
valid_batch_size: 8
test_batch_size: 8
lr: 1.0
ctc_weight: 0.5
sorting: ascending
dynamic_batching: false

# dynamic batching parameters, if used
dynamic_batch_sampler:
  feats_hop_size: 0.01
  max_batch_len: 20000 # in terms of frames
  shuffle_ex: true
  batch_ordering: random
  num_buckets: 20

# Feature parameters
sample_rate: 16000
n_fft: 400
n_mels: 40

opt_class: !name:torch.optim.Adadelta
  lr: 1.0
  rho: 0.95
  eps: 1.e-8

# Dataloader options
train_dataloader_opts:
  batch_size: 8

valid_dataloader_opts:
  batch_size: 8

test_dataloader_opts:
  batch_size: 8

# Model parameters
activation: &id001 !name:torch.nn.LeakyReLU
dropout: 0.15
cnn_blocks: 2
cnn_channels: (128, 256)
inter_layer_pooling_size: (2, 2)
cnn_kernelsize: (3, 3)
time_pooling_size: 4
rnn_class: &id002 !name:speechbrain.nnet.RNN.LSTM
rnn_layers: 4
rnn_neurons: 1024
rnn_bidirectional: true
dnn_blocks: 2
dnn_neurons: 512
emb_size: 128
dec_neurons: 1024
output_neurons: 1000 # Number of tokens (same as LM)
blank_index: 0
bos_index: 0
eos_index: 0

# Decoding parameters
min_decode_ratio: 0.0
max_decode_ratio: 1.0
valid_beam_size: 80
test_beam_size: 80
eos_threshold: 1.5
using_max_attn_shift: true
max_attn_shift: 240
lm_weight: 0.50
ctc_weight_decode: 0.0
coverage_penalty: 1.5
temperature: 1.25
temperature_lm: 1.25

epoch_counter: &id013 !new:speechbrain.utils.epoch_loop.EpochCounter

  limit: 25

normalize: &id008 !new:speechbrain.processing.features.InputNormalization
  norm_type: global

compute_features: !new:speechbrain.lobes.features.Fbank
  sample_rate: 16000
  n_fft: 400
  n_mels: 40

env_corrupt: &id009 !new:speechbrain.lobes.augment.EnvCorrupt
  openrir_folder: ../../cv_de_acc
  babble_prob: 0.0
  reverb_prob: 0.0
  noise_prob: 1.0
  noise_snr_low: 0
  noise_snr_high: 15

augmentation: !new:speechbrain.lobes.augment.TimeDomainSpecAugment
  sample_rate: 16000
  speeds: [95, 100, 105]

enc: &id003 !new:speechbrain.lobes.models.CRDNN.CRDNN
  input_shape: [null, null, 40]
  activation: *id001
  dropout: 0.15
  cnn_blocks: 2
  cnn_channels: (128, 256)
  cnn_kernelsize: (3, 3)
  inter_layer_pooling_size: (2, 2)
  time_pooling: true
  using_2d_pooling: false
  time_pooling_size: 4
  rnn_class: *id002
  rnn_layers: 4
  rnn_neurons: 1024
  rnn_bidirectional: true
  rnn_re_init: true
  dnn_blocks: 2
  dnn_neurons: 512
  use_rnnp: false

emb: &id004 !new:speechbrain.nnet.embedding.Embedding
  num_embeddings: 1000
  embedding_dim: 128

dec: &id005 !new:speechbrain.nnet.RNN.AttentionalRNNDecoder
  enc_dim: 512
  input_size: 128
  rnn_type: gru
  attn_type: location
  hidden_size: 1024
  attn_dim: 1024
  num_layers: 1
  scaling: 1.0
  channels: 10
  kernel_size: 100
  re_init: true
  dropout: 0.15

ctc_lin: &id006 !new:speechbrain.nnet.linear.Linear
  input_size: 512
  n_neurons: 1000

seq_lin: &id007 !new:speechbrain.nnet.linear.Linear
  input_size: 1024
  n_neurons: 1000

log_softmax: !new:speechbrain.nnet.activations.Softmax
  apply_log: true

ctc_cost: !name:speechbrain.nnet.losses.ctc_loss
  blank_index: 0

seq_cost: !name:speechbrain.nnet.losses.nll_loss
  label_smoothing: 0.1

# This is the RNNLM that is used according to the Huggingface repository
# NB: It has to match the pre-trained RNNLM!!
lm_model: &id010 !new:speechbrain.lobes.models.RNNLM.RNNLM

  output_neurons: 1000
  embedding_dim: 128
  activation: !name:torch.nn.LeakyReLU
  dropout: 0.0
  rnn_layers: 2
  rnn_neurons: 2048
  dnn_blocks: 1
  dnn_neurons: 512
  return_hidden: true # For inference

tokenizer: &id014 !new:sentencepiece.SentencePieceProcessor
                                                     # Models

modules:
  enc: *id003
  emb: *id004
  dec: *id005
  ctc_lin: *id006
  seq_lin: *id007
  normalize: *id008
  env_corrupt: *id009
  lm_model: *id010
model: &id011 !new:torch.nn.ModuleList
- [*id003, *id004, *id005, *id006, *id007]
valid_search: !new:speechbrain.decoders.S2SRNNBeamSearcher
  embedding: *id004
  decoder: *id005
  linear: *id007
  ctc_linear: *id006
  bos_index: 0
  eos_index: 0
  blank_index: 0
  min_decode_ratio: 0.0
  max_decode_ratio: 1.0
  beam_size: 80
  eos_threshold: 1.5
  using_max_attn_shift: true
  max_attn_shift: 240
  coverage_penalty: 1.5
  temperature: 1.25

test_search: !new:speechbrain.decoders.S2SRNNBeamSearchLM
  embedding: *id004
  decoder: *id005
  linear: *id007
  ctc_linear: *id006
  language_model: *id010
  bos_index: 0
  eos_index: 0
  blank_index: 0
  min_decode_ratio: 0.0
  max_decode_ratio: 1.0
  beam_size: 80
  eos_threshold: 1.5
  using_max_attn_shift: true
  max_attn_shift: 240
  coverage_penalty: 1.5
  lm_weight: 0.50
  ctc_weight: 0.0
  temperature: 1.25
  temperature_lm: 1.25

lr_annealing: &id012 !new:speechbrain.nnet.schedulers.NewBobScheduler
  initial_value: 1.0
  improvement_threshold: 0.0025
  annealing_factor: 0.8
  patient: 0

checkpointer: !new:speechbrain.utils.checkpoints.Checkpointer
  checkpoints_dir: results/CRDNN_BPE_960h_LM/1200/save
  recoverables:
    model: *id011
    scheduler: *id012
    normalizer: *id008
    counter: *id013
train_logger: !new:speechbrain.utils.train_logger.FileTrainLogger
  save_file: results/CRDNN_BPE_960h_LM/1200/train_log.txt

error_rate_computer: !name:speechbrain.utils.metric_stats.ErrorRateStats

cer_computer: !name:speechbrain.utils.metric_stats.ErrorRateStats
  split_tokens: true

# The pretrainer allows a mapping between pretrained files and instances that
# are declared in the yaml. E.g here, we will download the file lm.ckpt
# and it will be loaded into "lm" which is pointing to the <lm_model> defined
# before.
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
  collect_in: results/CRDNN_BPE_960h_LM/1200/save
  loadables:
    lm: *id010
    tokenizer: *id014
  paths:
    lm: ../../LM/results/RNN/2995/save/CKPT+2022-08-18+18-22-18+00/model.ckpt
    tokenizer: ../../Tokenizer/results/unigram//1000_unigram.model