respin_asru25_track4

ESPnet2 ASR model

saurabhk0322/respin_asru25_track4

This model was trained by wtc15 using respin_asru25 recipe in espnet.

Demo: How to use in ESPnet2

Follow the ESPnet installation instructions if you haven't done that already.

cd espnet

pip install -e .
cd egs2/respin_asru25/asr1
./run.sh --skip_data_prep false --skip_train true --download_model saurabhk0322/respin_asru25_track4

<!-- Generated by scripts/utils/show_asr_result.sh -->

RESULTS

Environments

• date: Wed Apr 23 02:53:40 IST 2025
• python version: 3.10.12 (main, Feb 4 2025, 14:57:36) [GCC 11.4.0]
• espnet version: espnet 202503
• pytorch version: pytorch 2.3.0+cu121
• Git hash: ``
  • Commit date: ``

exp/asru25/exp_large_lid_indic_char/asr_noaux_ssl_ml_con_e8_lin1024_bs6M_gacc1_ctc03_lid/decode_lid_asr_model_valid.acc.ave

WER

CER

TER

ASR config

<details><summary>expand</summary>

config: conf/tuning/train_asr_ssl_conformer_transformer_e8_linear1024_bs6M_gacc1_ml7-11.yaml
print_config: false
log_level: INFO
drop_last_iter: false
dry_run: false
iterator_type: sequence
valid_iterator_type: null
output_dir: exp/asru25/exp_large_lid_indic_char/asr_noaux_ssl_ml_con_e8_lin1024_bs6M_gacc1_ctc03_lid
ngpu: 1
seed: 2022
num_workers: 8
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
unused_parameters: false
sharded_ddp: false
use_deepspeed: false
deepspeed_config: null
gradient_as_bucket_view: true
ddp_comm_hook: null
cudnn_enabled: true
cudnn_benchmark: false
cudnn_deterministic: true
use_tf32: false
collect_stats: false
write_collected_feats: false
max_epoch: 70
patience: 5
val_scheduler_criterion:
- valid
- loss
early_stopping_criterion:
- valid
- loss
- min
best_model_criterion:
-   - valid
    - acc
    - max
keep_nbest_models: 5
nbest_averaging_interval: 0
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: true
log_interval: null
use_matplotlib: true
use_tensorboard: true
create_graph_in_tensorboard: false
use_wandb: false
wandb_project: null
wandb_id: null
wandb_entity: null
wandb_name: null
wandb_model_log_interval: -1
detect_anomaly: false
use_adapter: false
adapter: lora
save_strategy: all
adapter_conf: {}
pretrain_path: null
init_param: []
ignore_init_mismatch: false
freeze_param:
- frontend.upstream
num_iters_per_epoch: null
batch_size: 20
valid_batch_size: null
batch_bins: 9000000
valid_batch_bins: null
category_sample_size: 10
train_shape_file:
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/train/speech_shape
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/train/text_shape.char
valid_shape_file:
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/valid/speech_shape
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/valid/text_shape.char
batch_type: numel
valid_batch_type: null
fold_length:
- 80000
- 150
sort_in_batch: descending
shuffle_within_batch: false
sort_batch: descending
multiple_iterator: false
chunk_length: 500
chunk_shift_ratio: 0.5
num_cache_chunks: 1024
chunk_excluded_key_prefixes: []
chunk_default_fs: null
chunk_max_abs_length: null
chunk_discard_short_samples: true
train_data_path_and_name_and_type:
-   - dump/asru25_lid/raw/train_large_lid_sp/wav.scp
    - speech
    - sound
-   - dump/asru25_lid/raw/train_large_lid_sp/text
    - text
    - text
valid_data_path_and_name_and_type:
-   - dump/asru25_lid/raw/dev_lid/wav.scp
    - speech
    - sound
-   - dump/asru25_lid/raw/dev_lid/text
    - text
    - text
multi_task_dataset: false
allow_variable_data_keys: false
max_cache_size: 0.0
max_cache_fd: 32
allow_multi_rates: false
valid_max_cache_size: null
exclude_weight_decay: false
exclude_weight_decay_conf: {}
optim: adam
optim_conf:
    lr: 0.002
    weight_decay: 1.0e-06
scheduler: warmuplr
scheduler_conf:
    warmup_steps: 15000
token_list:
- <blank>
- <unk>
- <space>
- ा
- क
- े
- र
- ल
- न
- स
- ्
- त
- '['
- ']'
- म
- ी
- ि
- ್
- ह
- य
- ब
- प
- ो
- ్
- া
- व
- ज
- ं
- ు
- ా
- ి
- ಿ
- ে
- ಾ
- द
- র
- న
- ల
- ग
- ರ
- m
- ం
- క
- ক
- ర
- ು
- ನ
- ্
- ಕ
- ु
- च
- ट
- ತ
- ै
- ದ
- ি
- ख
- ೆ
- ಗ
- t
- ప
- త
- b
- h
- इ
- अ
- n
- आ
- ू
- ন
- ಸ
- ಲ
- వ
- ట
- స
- ವ
- ే
- छ
- ব
- ল
- ಯ
- भ
- श
- ಂ
- ಮ
- య
- য
- ध
- డ
- ద
- ए
- थ
- మ
- ಬ
- చ
- ण
- ड
- ম
- ई
- उ
- স
- ত
- ಳ
- ో
- ಡ
- फ
- g
- r
- e
- గ
- প
- ট
- য়
- c
- k
- ಟ
- ె
- .
- ಹ
- ಪ
- ೇ
- బ
- হ
- ু
- দ
- ष
- ো
- ీ
- জ
- ड़
- ಅ
- ೊ
- ই
- গ
- అ
- घ
- ಣ
- ठ
- ೋ
- চ
- ँ
- ొ
- ौ
- ছ
- ఎ
- ओ
- শ
- আ
- ూ
- జ
- ಜ
- থ
- ভ
- ಇ
- ೂ
- ಷ
- ಚ
- এ
- ষ
- ై
- ೀ
- ఉ
- ಎ
- ಆ
- ळ
- ধ
- ृ
- ী
- উ
- ফ
- খ
- ড
- ॉ
- ಶ
- অ
- ೈ
- ధ
- ং
- झ
- ఇ
- ఆ
- ష
- ढ
- ढ़
- భ
- శ
- ఏ
- ಧ
- ও
- ಒ
- ಭ
- ళ
- ಉ
- ॅ
- ಫ
- ऊ
- ఫ
- ಥ
- ऑ
- ణ
- ড়
- ণ
- ঙ
- ऋ
- ಖ
- ऽ
- హ
- థ
- औ
- ೃ
- ঁ
- ೌ
- ఒ
- ఖ
- ৃ
- ఈ
- ಏ
- ঠ
- ౌ
- ಐ
- ৈ
- ऐ
- ऱ
- ఐ
- ূ
- ञ
- ৎ
- ృ
- ज़
- ঞ
- ಈ
- ঘ
- ঋ
- ঝ
- फ़
- ৌ
- ಠ
- ः
- ఓ
- ಘ
- ಛ
- ಓ
- ఊ
- ఋ
- ಔ
- ఛ
- ಞ
- ॲ
- ಊ
- ఘ
- ঢ
- ख़
- ়
- ऍ
- ಋ
- क़
- ఠ
- ঢ়
- ঃ
- ़
- ೕ
- ఔ
- ಢ
- ঊ
- ఱ
- ಃ
- ಝ
- ङ
- ఢ
- ग़
- ఞ
- ঐ
- ঔ
- ॠ
- ':'
- ೯
- ೖ
- ঈ
- ః
- ౖ
- ৠ
- ౦
- <sos/eos>
init: null
input_size: null
ctc_conf:
    dropout_rate: 0.0
    ctc_type: builtin
    reduce: true
    ignore_nan_grad: null
    zero_infinity: true
    brctc_risk_strategy: exp
    brctc_group_strategy: end
    brctc_risk_factor: 0.0
joint_net_conf: null
use_preprocessor: true
use_lang_prompt: false
use_nlp_prompt: false
token_type: char
bpemodel: null
non_linguistic_symbols: data_respin/data_asru25/nlsyms.txt
cleaner: null
g2p: null
speech_volume_normalize: null
rir_scp: null
rir_apply_prob: 1.0
noise_scp: null
noise_apply_prob: 1.0
noise_db_range: '13_15'
short_noise_thres: 0.5
aux_ctc_tasks: []
frontend: s3prl
frontend_conf:
    frontend_conf:
        upstream: wav2vec2_url
        path_or_url: /home1/Saurabh/exp/espnet/RESPIN/DID_ASR/CTC_AUX/ckpts_converted/indicw2v_base_pretrained.pt
        normalize: false
    download_dir: /home1/Saurabh/exp/espnet/RESPIN/DID_ASR/CTC_AUX/ckpts_converted
    multilayer_feature: true
    fs: 16k
specaug: specaug
specaug_conf:
    apply_time_warp: true
    time_warp_window: 5
    time_warp_mode: bicubic
    apply_freq_mask: true
    freq_mask_width_range:
    - 0
    - 27
    num_freq_mask: 2
    apply_time_mask: true
    time_mask_width_ratio_range:
    - 0.0
    - 0.05
    num_time_mask: 5
normalize: utterance_mvn
normalize_conf: {}
model: espnet
model_conf:
    ctc_weight: 0.3
    lsm_weight: 0.1
    length_normalized_loss: false
preencoder: linear
preencoder_conf:
    input_size: 768
    output_size: 80
encoder: conformer
encoder_conf:
    output_size: 256
    attention_heads: 4
    linear_units: 1024
    num_blocks: 8
    dropout_rate: 0.1
    positional_dropout_rate: 0.1
    attention_dropout_rate: 0.1
    input_layer: conv2d2
    normalize_before: true
    macaron_style: true
    rel_pos_type: latest
    pos_enc_layer_type: rel_pos
    selfattention_layer_type: rel_selfattn
    activation_type: swish
    use_cnn_module: true
    cnn_module_kernel: 31
    use_flash_attn: false
postencoder: null
postencoder_conf: {}
decoder: transformer
decoder_conf:
    attention_heads: 4
    linear_units: 2048
    num_blocks: 6
    dropout_rate: 0.1
    positional_dropout_rate: 0.1
    self_attention_dropout_rate: 0.1
    src_attention_dropout_rate: 0.1
    layer_drop_rate: 0.0
    use_flash_attn: false
preprocessor: default
preprocessor_conf: {}
required:
- output_dir
- token_list
version: '202503'
distributed: false

</details>

Citing ESPnet

@inproceedings{watanabe2018espnet,
  author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
  title={{ESPnet}: End-to-End Speech Processing Toolkit},
  year={2018},
  booktitle={Proceedings of Interspeech},
  pages={2207--2211},
  doi={10.21437/Interspeech.2018-1456},
  url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
}

or arXiv:

@misc{watanabe2018espnet,
  title={ESPnet: End-to-End Speech Processing Toolkit},
  author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
  year={2018},
  eprint={1804.00015},
  archivePrefix={arXiv},
  primaryClass={cs.CL}
}