--- tags: - espnet - audio - speech-translation language: - en - es datasets: - cvss license: cc-by-4.0 --- ## ESPnet2 S2ST model Vocoder is located [here](https://huggingface.co/espnet/cvss-c_en_wavegan_hubert_vocoder), trained by [realzza](https://github.com/realzza) ### `espnet/jiyang_tang_cvss-c_es-en_discrete_unit` This model was trained by Jiyang Tang using cvss 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 c002f05ab3ef82938b6a980806cd7f97baba2299 pip install -e . cd egs2/cvss/s2st1 ./run.sh --skip_data_prep false --skip_train true --download_model espnet/jiyang_tang_cvss-c_es-en_discrete_unit ``` # RESULTS ## Environments - date: `Wed Oct 4 22:20:55 EDT 2023` - python version: `3.10.13 (main, Sep 11 2023, 13:44:35) [GCC 11.2.0]` - espnet version: `espnet 202308` - pytorch version: `pytorch 1.13.1` - Git hash: `79a3b3e2d9c9105f0f3f6d92d282e17f9ca91ed0` - Commit date: `Mon Sep 25 16:39:40 2023 -0400` ## S2ST config
expand ``` config: conf/train_s2st_discrete_unit.yaml print_config: false log_level: INFO drop_last_iter: false dry_run: false iterator_type: sequence valid_iterator_type: null output_dir: exp/s2st_train_s2st_discrete_unit_raw_fbank_es_en ngpu: 1 seed: 0 num_workers: 2 num_att_plot: 0 dist_backend: nccl dist_init_method: env:// dist_world_size: 2 dist_rank: 0 local_rank: 0 dist_master_addr: localhost dist_master_port: 56635 dist_launcher: null multiprocessing_distributed: true unused_parameters: false sharded_ddp: false cudnn_enabled: true cudnn_benchmark: false cudnn_deterministic: true collect_stats: false write_collected_feats: false max_epoch: 500 patience: null val_scheduler_criterion: - valid - loss early_stopping_criterion: - valid - loss - min best_model_criterion: - - valid - loss - min - - train - loss - min keep_nbest_models: 5 nbest_averaging_interval: 0 grad_clip: 1.0 grad_clip_type: 2.0 grad_noise: false accum_grad: 4 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: 110 valid_batch_size: null batch_bins: 1000000 valid_batch_bins: null train_shape_file: - exp/s2st_stats_raw_es_en/train/tgt_speech_shape - exp/s2st_stats_raw_es_en/train/src_speech_shape - exp/s2st_stats_raw_es_en/train/src_text_shape.char - exp/s2st_stats_raw_es_en/train/tgt_text_shape.char valid_shape_file: - exp/s2st_stats_raw_es_en/valid/src_speech_shape - exp/s2st_stats_raw_es_en/valid/tgt_speech_shape - exp/s2st_stats_raw_es_en/valid/tgt_text_shape.char - exp/s2st_stats_raw_es_en/valid/src_text_shape.char batch_type: sorted valid_batch_type: null fold_length: - 800 - 150 - 150 - 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_es/text.km.hubert_layer6_500.en.unique - tgt_speech - text - - dump/raw/train_es/wav.scp.es - src_speech - sound - - dump/raw/train_es/text.es - src_text - text - - dump/raw/train_es/text.en - tgt_text - text valid_data_path_and_name_and_type: - - dump/raw/dev_es/wav.scp.es - src_speech - sound - - dump/raw/dev_es/text.km.hubert_layer6_500.en.unique - tgt_speech - text - - dump/raw/dev_es/text.en - tgt_text - text - - dump/raw/dev_es/text.es - src_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: adamw optim_conf: lr: 0.0005 eps: 1.0e-06 scheduler: warmuplr scheduler_conf: warmup_steps: 10000 s2st_type: discrete_unit tgt_token_list: - - - - e - a - t - i - o - s - n - r - h - l - d - c - u - m - f - p - g - w - y - b - v - k - '''' - x - j - z - q - ñ - '-' - í - ó - á - é - ú - â - . - ʻ - ð - º - ə - ā - ̇ - '!' - þ - src_token_list: - - - - E - A - O - S - N - R - I - L - D - T - C - U - M - P - . - B - G - V - F - H - Y - ',' - ó - '"' - Q - í - J - á - Z - ñ - X - ú - K - '!' - '?' - W - é - ':' - '-' - ¿ - ¡ - Á - '''' - ; - ü - ’ - — - É - ö - ã - Ó - ‘ - ō - “ - â - – - Ú - ë - ä - _ - ā - ´ - ū - ― - ¨ - ø - ô - ê - æ - Í - ì - ć - Ñ - č - е - À - à - '`' - ゴ - ː - '|' - Ş - ‹ - › - Š - Č - ё - š - ï - … - ß - ř - ă - ʻ - ý - ° - ė - ò - ミ - 箱 - 消 - し - ム - ś - „ - Ś - ə - 鮨 - 鮓 - ł - Ö - û - · - Ä - ń - « - » - ذ - ه - ب - ي - ة - ṃ - ě - ‧ - İ - ı - Ø - î - ī - ț - Æ - Þ - Ϙ - ª - の - Е - ð - '=' - Ð - '&' - ž - ” - œ - unit_token_list: - '2' - '179' - '408' - '66' - '135' - '442' - '7' - '130' - '106' - '112' - '195' - '363' - '278' - '249' - '280' - '243' - '279' - '197' - '16' - '270' - '483' - '212' - '437' - '313' - '429' - '110' - '19' - '142' - '152' - '175' - '84' - '34' - '359' - '14' - '269' - '267' - '41' - '60' - '190' - '450' - '180' - '171' - '209' - '348' - '55' - '383' - '56' - '158' - '17' - '200' - '53' - '35' - '390' - '122' - '255' - '491' - '452' - '471' - '420' - '306' - '11' - '54' - '9' - '26' - '29' - '454' - '104' - '107' - '30' - '147' - '257' - '448' - '51' - '232' - '74' - '43' - '294' - '151' - '146' - '226' - '45' - '461' - '63' - '369' - '244' - '4' - '76' - '131' - '27' - '327' - '177' - '204' - '139' - '358' - '6' - '284' - '310' - '415' - '182' - '407' - '326' - '319' - '231' - '88' - '476' - '109' - '166' - '417' - '456' - '105' - '354' - '0' - '318' - '336' - '314' - '159' - '281' - '413' - '95' - '73' - '296' - '422' - '432' - '39' - '431' - '36' - '447' - '468' - '427' - '378' - '248' - '322' - '47' - '220' - '82' - '181' - '391' - '494' - '344' - '435' - '178' - '61' - '129' - '114' - '302' - '392' - '150' - '223' - '79' - '438' - '262' - '371' - '203' - '239' - '488' - '247' - '283' - '416' - '68' - '395' - '184' - '474' - '141' - '89' - '342' - '13' - '298' - '125' - '191' - '165' - '24' - '441' - '227' - '196' - '258' - '133' - '168' - '64' - '123' - '400' - '81' - '217' - '253' - '132' - '285' - '28' - '188' - '375' - '213' - '242' - '236' - '453' - '225' - '164' - '261' - '374' - '272' - '325' - '495' - '460' - '330' - '48' - '451' - '323' - '458' - '263' - '144' - '160' - '149' - '406' - '77' - '33' - '368' - '332' - '205' - '50' - '290' - '401' - '490' - '331' - '436' - '5' - '186' - '288' - '148' - '219' - '215' - '93' - '434' - '103' - '489' - '21' - '92' - '386' - '97' - '328' - '121' - '301' - '46' - '224' - '154' - '80' - '352' - '96' - '124' - '69' - '457' - '83' - '52' - '85' - '62' - '380' - '410' - '167' - '333' - '31' - '315' - '78' - '271' - '10' - '492' - '49' - '208' - '295' - '86' - '199' - '445' - '140' - '357' - '187' - '161' - '238' - '351' - '155' - '193' - '345' - '486' - '37' - '266' - '185' - '143' - '361' - '174' - '430' - '349' - '23' - '423' - '388' - '309' - '470' - '169' - '370' - '463' - '245' - '320' - '237' - '316' - '277' - '482' - '218' - '198' - '117' - '428' - '340' - '475' - '418' - '275' - '299' - '297' - '90' - '260' - '276' - '137' - '366' - '353' - '341' - '241' - '496' - '228' - '287' - '214' - '264' - '108' - '44' - '350' - '3' - '286' - '303' - '12' - '372' - '156' - '321' - '116' - '385' - '194' - '360' - '119' - '145' - '22' - '414' - '462' - '70' - '449' - '251' - '387' - '466' - '273' - '440' - '58' - '304' - '162' - '404' - '15' - '176' - '384' - '293' - '397' - '173' - '59' - '485' - '75' - '102' - '282' - '233' - '115' - '210' - '222' - '18' - '498' - '99' - '398' - '91' - '221' - '396' - '479' - '300' - '339' - '367' - '459' - '20' - '216' - '426' - '87' - '382' - '424' - '446' - '1' - '265' - '172' - '346' - '259' - '183' - '113' - '307' - '311' - '201' - '170' - '240' - '25' - '291' - '393' - '444' - '292' - '334' - '234' - '379' - '153' - '42' - '250' - '409' - '464' - '394' - '256' - '399' - '465' - '381' - '364' - '157' - '356' - '268' - '65' - '343' - '338' - '493' - '100' - '405' - '421' - '111' - '289' - '365' - '246' - '8' - '101' - '163' - '252' - '138' - '72' - '373' - '362' - '120' - '425' - '480' - '32' - '254' - '202' - '484' - '412' - '473' - '71' - '355' - '443' - '134' - '324' - '118' - '402' - '230' - '67' - '98' - '335' - '317' - '57' - '329' - '229' - '419' - '94' - '128' - '376' - '433' - '192' - '235' - '38' - '312' - '347' - '499' - '274' - '389' - '127' - '439' - '207' - '478' - '403' - '467' - '411' - '455' - '337' - '469' - '206' - '497' - '136' - '481' - '487' - '40' - '477' - '472' - '189' - '308' - '377' - '305' - '211' - '126' - - odim: null init: null input_size: null output_size: 500 asr_ctc: true st_ctc: true asr_ctc_conf: dropout_rate: 0.0 ctc_type: builtin reduce: true ignore_nan_grad: null zero_infinity: true st_ctc_conf: dropout_rate: 0.0 ctc_type: builtin reduce: true ignore_nan_grad: null zero_infinity: true model_conf: ignore_id: -1 report_cer: true report_wer: true report_bleu: true sym_space: sym_blank: extract_feats_in_collect_stats: true use_preprocessor: true tgt_token_type: char src_token_type: char tgt_bpemodel: null src_bpemodel: null non_linguistic_symbols: null cleaner: null tgt_g2p: null src_g2p: null losses: - name: asr_ctc type: ctc conf: weight: 1.6 - name: src_attn type: attention conf: weight: 8.0 smoothing: 0.2 padding_idx: -1 - name: tgt_attn type: attention conf: weight: 8.0 smoothing: 0.2 padding_idx: -1 - name: st_ctc type: ctc conf: weight: 1.6 - name: synthesis type: attention conf: weight: 1.6 smoothing: 0.2 padding_idx: -1 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 frontend: default frontend_conf: n_fft: 512 win_length: 400 hop_length: 160 fs: 16k tgt_feats_extract: null tgt_feats_extract_conf: {} 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: 10 src_normalize: global_mvn src_normalize_conf: stats_file: exp/s2st_stats_raw_es_en/train/src_feats_stats.npz tgt_normalize: utterance_mvn tgt_normalize_conf: {} preencoder: null preencoder_conf: {} encoder: transformer 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 normalize_before: true postencoder: null postencoder_conf: {} asr_decoder: transformer asr_decoder_conf: input_layer: embed num_blocks: 2 linear_units: 2048 attention_heads: 4 st_decoder: transformer st_decoder_conf: input_layer: embed num_blocks: 2 linear_units: 2048 attention_heads: 4 aux_attention: null aux_attention_conf: {} unit_encoder: null unit_encoder_conf: {} synthesizer: discrete_unit synthesizer_conf: input_layer: embed num_blocks: 6 linear_units: 2048 attention_heads: 8 loss: tacotron loss_conf: {} required: - output_dir version: '202308' distributed: true ```
### 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} } ```