ESPnet2 pretrained model, coppersj/yesno_asr_train_asr_raw_char_valid.acc.ave, fs=16k, lang=en

This model was trained by coppersj using yesno recipe in espnet.

• Python API

  See https://github.com/espnet/espnet_model_zoo

• Evaluate in the recipe

  git clone https://github.com/espnet/espnet
  cd espnet
  pip install -e .
  cd egs2/yesno/asr1
  ./run.sh --skip_data_prep false --skip_train true --download_model coppersj/yesno_asr_train_asr_raw_char_valid.acc.ave
  

• Results

  
  # RESULTS
  ## Environments
  - date: `Mon Sep 21 06:02:39 UTC 2020`
  - python version: `3.6.9 (default, Jul 17 2020, 12:50:27)  [GCC 8.4.0]`
  - espnet version: `espnet 0.9.3`
  - pytorch version: `pytorch 1.6.0+cu101`
  - Git hash: `624fdc847222b48dd0d34aef44f9e0cbec05756b`
    - Commit date: `Sun Sep 20 14:22:37 2020 +0000`
  
  ## asr_train_asr_raw_char
  ### WER
  
  |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
  |---|---|---|---|---|---|---|---|---|
  |decode_asr_model_valid.acc.ave/test_yesno|30|240|39.6|60.4|0.0|371.7|432.1|100.0|
  |decode_asr_model_valid.acc.ave/train_dev|2|16|62.5|37.5|0.0|375.0|412.5|100.0|
  
  ### CER
  
  |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
  |---|---|---|---|---|---|---|---|---|
  |decode_asr_model_valid.acc.ave/test_yesno|30|835|47.9|52.1|0.0|303.0|355.1|100.0|
  |decode_asr_model_valid.acc.ave/train_dev|2|52|65.4|34.6|0.0|334.6|369.2|100.0|

• ASR config

  config: conf/train_asr.yaml
  print_config: false
  log_level: INFO
  dry_run: false
  iterator_type: sequence
  output_dir: exp/asr_train_asr_raw_char
  ngpu: 1
  seed: 0
  num_workers: 1
  num_att_plot: 3
  dist_backend: nccl
  dist_init_method: env://
  dist_world_size: null
  dist_rank: null
  local_rank: 0
  dist_master_addr: null
  dist_master_port: null
  dist_launcher: null
  multiprocessing_distributed: false
  cudnn_enabled: true
  cudnn_benchmark: false
  cudnn_deterministic: true
  collect_stats: false
  write_collected_feats: false
  max_epoch: 20
  patience: 3
  val_scheduler_criterion:
  - valid
  - acc
  early_stopping_criterion:
  - valid
  - acc
  - max
  best_model_criterion:
  -   - valid
      - acc
      - max
  keep_nbest_models: 1
  grad_clip: 5.0
  grad_clip_type: 2.0
  grad_noise: false
  accum_grad: 1
  no_forward_run: false
  resume: true
  train_dtype: float32
  use_amp: false
  log_interval: null
  pretrain_path: []
  pretrain_key: []
  num_iters_per_epoch: null
  batch_size: 5
  valid_batch_size: null
  batch_bins: 1000000
  valid_batch_bins: null
  train_shape_file:
  - exp/asr_stats_raw/train/speech_shape
  - exp/asr_stats_raw/train/text_shape.char
  valid_shape_file:
  - exp/asr_stats_raw/valid/speech_shape
  - exp/asr_stats_raw/valid/text_shape.char
  batch_type: folded
  valid_batch_type: null
  fold_length:
  - 80000
  - 150
  sort_in_batch: descending
  sort_batch: descending
  multiple_iterator: false
  chunk_length: 500
  chunk_shift_ratio: 0.5
  num_cache_chunks: 1024
  train_data_path_and_name_and_type:
  -   - dump/raw/train_nodev/wav.scp
      - speech
      - sound
  -   - dump/raw/train_nodev/text
      - text
      - text
  valid_data_path_and_name_and_type:
  -   - dump/raw/train_dev/wav.scp
      - speech
      - sound
  -   - dump/raw/train_dev/text
      - text
      - text
  allow_variable_data_keys: false
  max_cache_size: 0.0
  valid_max_cache_size: null
  optim: adadelta
  optim_conf:
      lr: 1.0
      rho: 0.95
      eps: 1.0e-08
      weight_decay: 0
  scheduler: null
  scheduler_conf: {}
  token_list:
  - 
  - 
  - 
  - N
  - O
  - Y
  - E
  - S
  - 
  init: chainer
  input_size: null
  ctc_conf:
      dropout_rate: 0.0
      ctc_type: builtin
      reduce: true
  model_conf:
      ctc_weight: 0.5
  use_preprocessor: true
  token_type: char
  bpemodel: null
  non_linguistic_symbols: null
  cleaner: null
  g2p: null
  frontend: default
  frontend_conf:
      fs: 16k
  specaug: null
  specaug_conf: {}
  normalize: global_mvn
  normalize_conf:
      stats_file: exp/asr_stats_raw/train/feats_stats.npz
  encoder: rnn
  encoder_conf:
      rnn_type: lstm
      bidirectional: true
      use_projection: true
      num_layers: 3
      hidden_size: 128
      output_size: 128
  decoder: rnn
  decoder_conf:
      rnn_type: lstm
      num_layers: 1
      hidden_size: 128
      sampling_probability: 0.0
      att_conf:
          atype: location
          adim: 128
          aconv_chans: 10
          aconv_filts: 100
  required:
  - output_dir
  - token_list
  distributed: false

• LM config

  NONE