--- tags: - espnet - audio - automatic-speech-recognition language: eu datasets: - commonvoice license: cc-by-4.0 --- ## ESPnet2 ASR model ### `espnet/zuazo_commonvoice_asr_train_asr_conformer5_raw_eu_bpe150_sp` This model was trained by Xabier de Zuazo using commonvoice recipe in [espnet](https://github.com/espnet/espnet/). ### Demo: How to use in ESPnet2 Follow the [ESPnet installation instructions](https://espnet.github.io/espnet/installation.html) if you haven't done that already. ```bash cd espnet git checkout 5d0758e2a7063b82d1f10a8ac2de98eb6cf8a352 pip install -e . cd egs2/commonvoice/asr1.conformer.lm.best ./run.sh --skip_data_prep false --skip_train true --download_model espnet/zuazo_commonvoice_asr_train_asr_conformer5_raw_eu_bpe150_sp ``` # RESULTS ## Environments - date: `Thu Sep 21 09:55:45 CEST 2023` - python version: `3.8.17 | packaged by conda-forge | (default, Jun 16 2023, 07:06:00) [GCC 11.4.0]` - espnet version: `espnet 202308` - pytorch version: `pytorch 2.0.1` - Git hash: `5d0758e2a7063b82d1f10a8ac2de98eb6cf8a352` - Commit date: `Wed Aug 30 18:03:42 2023 -0400` ## exp/asr_train_asr_conformer5_raw_eu_bpe150_sp ### WER |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err| |---|---|---|---|---|---|---|---|---| |decode_asr_lm_lm_train_lm_eu_bpe150_valid.loss.ave_asr_model_valid.acc.ave/test_eu|6640|49267|92.8|6.8|0.4|0.8|8.0|33.3| ### CER |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err| |---|---|---|---|---|---|---|---|---| |decode_asr_lm_lm_train_lm_eu_bpe150_valid.loss.ave_asr_model_valid.acc.ave/test_eu|6640|373913|98.8|0.6|0.7|0.4|1.6|33.3| ### TER |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err| |---|---|---|---|---|---|---|---|---| |decode_asr_lm_lm_train_lm_eu_bpe150_valid.loss.ave_asr_model_valid.acc.ave/test_eu|6640|208360|97.4|1.5|1.1|0.5|3.1|33.3| ## exp/asr_train_asr_conformer5_raw_eu_bpe150_sp/decode_asr_lm_lm_train_lm_eu_bpe150_valid.loss.ave_asr_model_valid.acc.ave ### WER |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err| |---|---|---|---|---|---|---|---|---| |org/dev_eu|6640|49505|93.5|6.2|0.3|0.8|7.3|31.0| ### CER |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err| |---|---|---|---|---|---|---|---|---| |org/dev_eu|6640|376502|99.0|0.5|0.5|0.3|1.4|31.0| ### TER |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err| |---|---|---|---|---|---|---|---|---| |org/dev_eu|6640|209465|97.7|1.4|1.0|0.4|2.7|31.0| ## ASR config
expand ``` config: conf/tuning/train_asr_conformer5.yaml print_config: false log_level: INFO drop_last_iter: false dry_run: false iterator_type: sequence valid_iterator_type: null output_dir: exp/asr_train_asr_conformer5_raw_eu_bpe150_sp 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 unused_parameters: false sharded_ddp: false cudnn_enabled: true cudnn_benchmark: false cudnn_deterministic: true collect_stats: false write_collected_feats: false max_epoch: 50 patience: null val_scheduler_criterion: - valid - loss early_stopping_criterion: - valid - loss - min best_model_criterion: - - valid - acc - max keep_nbest_models: 10 nbest_averaging_interval: 0 grad_clip: 3 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 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 pretrain_path: null init_param: [] ignore_init_mismatch: false freeze_param: [] num_iters_per_epoch: null batch_size: 20 valid_batch_size: null batch_bins: 10000000 valid_batch_bins: null train_shape_file: - exp/asr_stats_raw_eu_bpe150_sp/train/speech_shape - exp/asr_stats_raw_eu_bpe150_sp/train/text_shape.bpe valid_shape_file: - exp/asr_stats_raw_eu_bpe150_sp/valid/speech_shape - exp/asr_stats_raw_eu_bpe150_sp/valid/text_shape.bpe 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: [] train_data_path_and_name_and_type: - - dump/raw/train_eu_sp/wav.scp - speech - sound - - dump/raw/train_eu_sp/text - text - text valid_data_path_and_name_and_type: - - dump/raw/dev_eu/wav.scp - speech - sound - - dump/raw/dev_eu/text - text - text allow_variable_data_keys: false max_cache_size: 0.0 max_cache_fd: 32 valid_max_cache_size: null exclude_weight_decay: false exclude_weight_decay_conf: {} optim: adam optim_conf: lr: 4.0 scheduler: noamlr scheduler_conf: model_size: 256 warmup_steps: 25000 token_list: - - - A - ▁ - I - E - Z - . - R - N - U - S - O - T - KO - K - ▁E - TU - TE - RA - EN - L - ',' - LA - TA - AK - ▁A - AN - ▁DA - RE - KA - P - GO - IN - B - M - ▁DU - RI - GU - ▁ETA - D - ER - UR - ▁BA - ▁P - H - MA - ▁G - ▁I - ▁HA - TZEN - LE - ▁EZ - ▁O - EK - GI - ▁BAT - DA - DU - TZA - KI - DI - RO - ▁GA - REN - AR - TEN - GA - TIK - RRI - ▁BI - LI - ▁BER - G - ▁AR - TO - ERA - AREN - ▁ZI - ▁DE - ▁BE - X - BA - ▁DI - ▁IZAN - ▁ZE - ETAN - ▁ZEN - EAN - IA - ▁JA - ▁ERE - ▁DITU - ▁ZA - ▁ERA - LO - ▁HOR - NTZ - ▁DIRA - MEN - ▁HI - ▁F - F - LDE - ZIO - '?' - ▁ZU - '-' - DO - ▁EGIN - TZEKO - ▁BEHAR - TZI - BIL - ▁IN - RIK - ▁HORI - ▁SA - ▁NA - BIDE - ▁KON - ▁HE - ▁ZUEN - ▁MU - ▁BESTE - ▁SO - ▁HERRI - ▁IKAS - ▁NO - ▁ALD - ▁NI - ▁TX - ABE - KETA - ▁BAINA - C - '!' - V - Y - ':' - ; - '"' - í - Q - ñ - W - J - ‘ - ’ - init: null input_size: null ctc_conf: dropout_rate: 0.0 ctc_type: builtin reduce: true ignore_nan_grad: null zero_infinity: true joint_net_conf: null use_preprocessor: true token_type: bpe bpemodel: data/eu_token_list/bpe_unigram150/bpe.model non_linguistic_symbols: null 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: default frontend_conf: n_fft: 512 win_length: 400 hop_length: 160 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 - 30 num_freq_mask: 2 apply_time_mask: true time_mask_width_range: - 0 - 40 num_time_mask: 2 normalize: global_mvn normalize_conf: stats_file: exp/asr_stats_raw_eu_bpe150_sp/train/feats_stats.npz model: espnet model_conf: ctc_weight: 0.3 lsm_weight: 0.1 length_normalized_loss: false preencoder: null preencoder_conf: {} encoder: conformer encoder_conf: input_layer: conv2d num_blocks: 12 linear_units: 2048 dropout_rate: 0.1 output_size: 256 attention_heads: 4 attention_dropout_rate: 0.0 pos_enc_layer_type: rel_pos selfattention_layer_type: rel_selfattn activation_type: swish macaron_style: true use_cnn_module: true cnn_module_kernel: 15 postencoder: null postencoder_conf: {} decoder: transformer decoder_conf: input_layer: embed num_blocks: 6 linear_units: 2048 dropout_rate: 0.1 preprocessor: default preprocessor_conf: {} required: - output_dir - token_list version: '202308' distributed: false ```
### Citing ESPnet ```BibTex @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: ```bibtex @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} } ```