PolyAI/minds14
MInDS-14 MINDS-14 is training and evaluation resource for intent detection task with spoken data. It covers 14 intents extracted from a commercial system in the e-banking domain, associated with spoken examples in 14 diverse language varieties. Example MInDS-14 can be downloaded and used as follows: from datasets import load_dataset minds_14 = load_dataset("PolyAI/minds14", "fr-FR") # for French # to download all data for multi-lingual fine-tuning uncomment… See the full description on the dataset page: https://huggingface.co/datasets/PolyAI/minds14.
10815k
1---2annotations_creators:3- expert-generated4- crowdsourced5- machine-generated6language_creators:7- crowdsourced8- expert-generated9language:10- en11- fr12- it13- es14- pt15- de16- nl17- ru18- pl19- cs20- ko21- zh22license:23- cc-by-4.024multilinguality:25- multilingual26size_categories:27- 10K<n<100K28task_categories:29- automatic-speech-recognition30task_ids:31- keyword-spotting32pretty_name: MInDS-1433language_bcp47:34- en35- en-GB36- en-US37- en-AU38- fr39- it40- es41- pt42- de43- nl44- ru45- pl46- cs47- ko48- zh49tags:50- speech-recognition51configs:52- config_name: all53 data_files:54 - split: train55 path: all/train-*56- config_name: cs-CZ57 data_files:58 - split: train59 path: cs-CZ/train-*60- config_name: de-DE61 data_files:62 - split: train63 path: de-DE/train-*64- config_name: en-AU65 data_files:66 - split: train67 path: en-AU/train-*68- config_name: en-GB69 data_files:70 - split: train71 path: en-GB/train-*72- config_name: en-US73 data_files:74 - split: train75 path: en-US/train-*76- config_name: es-ES77 data_files:78 - split: train79 path: es-ES/train-*80- config_name: fr-FR81 data_files:82 - split: train83 path: fr-FR/train-*84- config_name: it-IT85 data_files:86 - split: train87 path: it-IT/train-*88- config_name: ko-KR89 data_files:90 - split: train91 path: ko-KR/train-*92- config_name: nl-NL93 data_files:94 - split: train95 path: nl-NL/train-*96- config_name: pl-PL97 data_files:98 - split: train99 path: pl-PL/train-*100- config_name: pt-PT101 data_files:102 - split: train103 path: pt-PT/train-*104- config_name: ru-RU105 data_files:106 - split: train107 path: ru-RU/train-*108- config_name: zh-CN109 data_files:110 - split: train111 path: zh-CN/train-*112dataset_info:113- config_name: all114 features:115 - name: path116 dtype: string117 - name: audio118 dtype:119 audio:120 sampling_rate: 8000121 - name: transcription122 dtype: string123 - name: english_transcription124 dtype: string125 - name: intent_class126 dtype:127 class_label:128 names:129 '0': abroad130 '1': address131 '2': app_error132 '3': atm_limit133 '4': balance134 '5': business_loan135 '6': card_issues136 '7': cash_deposit137 '8': direct_debit138 '9': freeze139 '10': high_value_payment140 '11': joint_account141 '12': latest_transactions142 '13': pay_bill143 - name: lang_id144 dtype:145 class_label:146 names:147 '0': cs-CZ148 '1': de-DE149 '2': en-AU150 '3': en-GB151 '4': en-US152 '5': es-ES153 '6': fr-FR154 '7': it-IT155 '8': ko-KR156 '9': nl-NL157 '10': pl-PL158 '11': pt-PT159 '12': ru-RU160 '13': zh-CN161 splits:162 - name: train163 num_bytes: 628290192.68164 num_examples: 8168165 download_size: 566674935166 dataset_size: 628290192.68167- config_name: cs-CZ168 features:169 - name: path170 dtype: string171 - name: audio172 dtype:173 audio:174 sampling_rate: 8000175 - name: transcription176 dtype: string177 - name: english_transcription178 dtype: string179 - name: intent_class180 dtype:181 class_label:182 names:183 '0': abroad184 '1': address185 '2': app_error186 '3': atm_limit187 '4': balance188 '5': business_loan189 '6': card_issues190 '7': cash_deposit191 '8': direct_debit192 '9': freeze193 '10': high_value_payment194 '11': joint_account195 '12': latest_transactions196 '13': pay_bill197 - name: lang_id198 dtype:199 class_label:200 names:201 '0': cs-CZ202 '1': de-DE203 '2': en-AU204 '3': en-GB205 '4': en-US206 '5': es-ES207 '6': fr-FR208 '7': it-IT209 '8': ko-KR210 '9': nl-NL211 '10': pl-PL212 '11': pt-PT213 '12': ru-RU214 '13': zh-CN215 splits:216 - name: train217 num_bytes: 40369391.0218 num_examples: 574219 download_size: 36473997220 dataset_size: 40369391.0221- config_name: de-DE222 features:223 - name: path224 dtype: string225 - name: audio226 dtype:227 audio:228 sampling_rate: 8000229 - name: transcription230 dtype: string231 - name: english_transcription232 dtype: string233 - name: intent_class234 dtype:235 class_label:236 names:237 '0': abroad238 '1': address239 '2': app_error240 '3': atm_limit241 '4': balance242 '5': business_loan243 '6': card_issues244 '7': cash_deposit245 '8': direct_debit246 '9': freeze247 '10': high_value_payment248 '11': joint_account249 '12': latest_transactions250 '13': pay_bill251 - name: lang_id252 dtype:253 class_label:254 names:255 '0': cs-CZ256 '1': de-DE257 '2': en-AU258 '3': en-GB259 '4': en-US260 '5': es-ES261 '6': fr-FR262 '7': it-IT263 '8': ko-KR264 '9': nl-NL265 '10': pl-PL266 '11': pt-PT267 '12': ru-RU268 '13': zh-CN269 splits:270 - name: train271 num_bytes: 53448213.0272 num_examples: 611273 download_size: 41073246274 dataset_size: 53448213.0275- config_name: en-AU276 features:277 - name: path278 dtype: string279 - name: audio280 dtype:281 audio:282 sampling_rate: 8000283 - name: transcription284 dtype: string285 - name: english_transcription286 dtype: string287 - name: intent_class288 dtype:289 class_label:290 names:291 '0': abroad292 '1': address293 '2': app_error294 '3': atm_limit295 '4': balance296 '5': business_loan297 '6': card_issues298 '7': cash_deposit299 '8': direct_debit300 '9': freeze301 '10': high_value_payment302 '11': joint_account303 '12': latest_transactions304 '13': pay_bill305 - name: lang_id306 dtype:307 class_label:308 names:309 '0': cs-CZ310 '1': de-DE311 '2': en-AU312 '3': en-GB313 '4': en-US314 '5': es-ES315 '6': fr-FR316 '7': it-IT317 '8': ko-KR318 '9': nl-NL319 '10': pl-PL320 '11': pt-PT321 '12': ru-RU322 '13': zh-CN323 splits:324 - name: train325 num_bytes: 44154982.0326 num_examples: 654327 download_size: 37348052328 dataset_size: 44154982.0329- config_name: en-GB330 features:331 - name: path332 dtype: string333 - name: audio334 dtype:335 audio:336 sampling_rate: 8000337 - name: transcription338 dtype: string339 - name: english_transcription340 dtype: string341 - name: intent_class342 dtype:343 class_label:344 names:345 '0': abroad346 '1': address347 '2': app_error348 '3': atm_limit349 '4': balance350 '5': business_loan351 '6': card_issues352 '7': cash_deposit353 '8': direct_debit354 '9': freeze355 '10': high_value_payment356 '11': joint_account357 '12': latest_transactions358 '13': pay_bill359 - name: lang_id360 dtype:361 class_label:362 names:363 '0': cs-CZ364 '1': de-DE365 '2': en-AU366 '3': en-GB367 '4': en-US368 '5': es-ES369 '6': fr-FR370 '7': it-IT371 '8': ko-KR372 '9': nl-NL373 '10': pl-PL374 '11': pt-PT375 '12': ru-RU376 '13': zh-CN377 splits:378 - name: train379 num_bytes: 39143835.0380 num_examples: 592381 download_size: 34551079382 dataset_size: 39143835.0383- config_name: en-US384 features:385 - name: path386 dtype: string387 - name: audio388 dtype:389 audio:390 sampling_rate: 8000391 - name: transcription392 dtype: string393 - name: english_transcription394 dtype: string395 - name: intent_class396 dtype:397 class_label:398 names:399 '0': abroad400 '1': address401 '2': app_error402 '3': atm_limit403 '4': balance404 '5': business_loan405 '6': card_issues406 '7': cash_deposit407 '8': direct_debit408 '9': freeze409 '10': high_value_payment410 '11': joint_account411 '12': latest_transactions412 '13': pay_bill413 - name: lang_id414 dtype:415 class_label:416 names:417 '0': cs-CZ418 '1': de-DE419 '2': en-AU420 '3': en-GB421 '4': en-US422 '5': es-ES423 '6': fr-FR424 '7': it-IT425 '8': ko-KR426 '9': nl-NL427 '10': pl-PL428 '11': pt-PT429 '12': ru-RU430 '13': zh-CN431 splits:432 - name: train433 num_bytes: 38787013.0434 num_examples: 563435 download_size: 34196221436 dataset_size: 38787013.0437- config_name: es-ES438 features:439 - name: path440 dtype: string441 - name: audio442 dtype:443 audio:444 sampling_rate: 8000445 - name: transcription446 dtype: string447 - name: english_transcription448 dtype: string449 - name: intent_class450 dtype:451 class_label:452 names:453 '0': abroad454 '1': address455 '2': app_error456 '3': atm_limit457 '4': balance458 '5': business_loan459 '6': card_issues460 '7': cash_deposit461 '8': direct_debit462 '9': freeze463 '10': high_value_payment464 '11': joint_account465 '12': latest_transactions466 '13': pay_bill467 - name: lang_id468 dtype:469 class_label:470 names:471 '0': cs-CZ472 '1': de-DE473 '2': en-AU474 '3': en-GB475 '4': en-US476 '5': es-ES477 '6': fr-FR478 '7': it-IT479 '8': ko-KR480 '9': nl-NL481 '10': pl-PL482 '11': pt-PT483 '12': ru-RU484 '13': zh-CN485 splits:486 - name: train487 num_bytes: 44143874.0488 num_examples: 486489 download_size: 39069577490 dataset_size: 44143874.0491- config_name: fr-FR492 features:493 - name: path494 dtype: string495 - name: audio496 dtype:497 audio:498 sampling_rate: 8000499 - name: transcription500 dtype: string501 - name: english_transcription502 dtype: string503 - name: intent_class504 dtype:505 class_label:506 names:507 '0': abroad508 '1': address509 '2': app_error510 '3': atm_limit511 '4': balance512 '5': business_loan513 '6': card_issues514 '7': cash_deposit515 '8': direct_debit516 '9': freeze517 '10': high_value_payment518 '11': joint_account519 '12': latest_transactions520 '13': pay_bill521 - name: lang_id522 dtype:523 class_label:524 names:525 '0': cs-CZ526 '1': de-DE527 '2': en-AU528 '3': en-GB529 '4': en-US530 '5': es-ES531 '6': fr-FR532 '7': it-IT533 '8': ko-KR534 '9': nl-NL535 '10': pl-PL536 '11': pt-PT537 '12': ru-RU538 '13': zh-CN539 splits:540 - name: train541 num_bytes: 36200685.0542 num_examples: 539543 download_size: 32613161544 dataset_size: 36200685.0545- config_name: it-IT546 features:547 - name: path548 dtype: string549 - name: audio550 dtype:551 audio:552 sampling_rate: 8000553 - name: transcription554 dtype: string555 - name: english_transcription556 dtype: string557 - name: intent_class558 dtype:559 class_label:560 names:561 '0': abroad562 '1': address563 '2': app_error564 '3': atm_limit565 '4': balance566 '5': business_loan567 '6': card_issues568 '7': cash_deposit569 '8': direct_debit570 '9': freeze571 '10': high_value_payment572 '11': joint_account573 '12': latest_transactions574 '13': pay_bill575 - name: lang_id576 dtype:577 class_label:578 names:579 '0': cs-CZ580 '1': de-DE581 '2': en-AU582 '3': en-GB583 '4': en-US584 '5': es-ES585 '6': fr-FR586 '7': it-IT587 '8': ko-KR588 '9': nl-NL589 '10': pl-PL590 '11': pt-PT591 '12': ru-RU592 '13': zh-CN593 splits:594 - name: train595 num_bytes: 82852403.0596 num_examples: 696597 download_size: 59299376598 dataset_size: 82852403.0599- config_name: ko-KR600 features:601 - name: path602 dtype: string603 - name: audio604 dtype:605 audio:606 sampling_rate: 8000607 - name: transcription608 dtype: string609 - name: english_transcription610 dtype: string611 - name: intent_class612 dtype:613 class_label:614 names:615 '0': abroad616 '1': address617 '2': app_error618 '3': atm_limit619 '4': balance620 '5': business_loan621 '6': card_issues622 '7': cash_deposit623 '8': direct_debit624 '9': freeze625 '10': high_value_payment626 '11': joint_account627 '12': latest_transactions628 '13': pay_bill629 - name: lang_id630 dtype:631 class_label:632 names:633 '0': cs-CZ634 '1': de-DE635 '2': en-AU636 '3': en-GB637 '4': en-US638 '5': es-ES639 '6': fr-FR640 '7': it-IT641 '8': ko-KR642 '9': nl-NL643 '10': pl-PL644 '11': pt-PT645 '12': ru-RU646 '13': zh-CN647 splits:648 - name: train649 num_bytes: 39733949.0650 num_examples: 592651 download_size: 34163604652 dataset_size: 39733949.0653- config_name: nl-NL654 features:655 - name: path656 dtype: string657 - name: audio658 dtype:659 audio:660 sampling_rate: 8000661 - name: transcription662 dtype: string663 - name: english_transcription664 dtype: string665 - name: intent_class666 dtype:667 class_label:668 names:669 '0': abroad670 '1': address671 '2': app_error672 '3': atm_limit673 '4': balance674 '5': business_loan675 '6': card_issues676 '7': cash_deposit677 '8': direct_debit678 '9': freeze679 '10': high_value_payment680 '11': joint_account681 '12': latest_transactions682 '13': pay_bill683 - name: lang_id684 dtype:685 class_label:686 names:687 '0': cs-CZ688 '1': de-DE689 '2': en-AU690 '3': en-GB691 '4': en-US692 '5': es-ES693 '6': fr-FR694 '7': it-IT695 '8': ko-KR696 '9': nl-NL697 '10': pl-PL698 '11': pt-PT699 '12': ru-RU700 '13': zh-CN701 splits:702 - name: train703 num_bytes: 51418818.0704 num_examples: 654705 download_size: 46536858706 dataset_size: 51418818.0707- config_name: pl-PL708 features:709 - name: path710 dtype: string711 - name: audio712 dtype:713 audio:714 sampling_rate: 8000715 - name: transcription716 dtype: string717 - name: english_transcription718 dtype: string719 - name: intent_class720 dtype:721 class_label:722 names:723 '0': abroad724 '1': address725 '2': app_error726 '3': atm_limit727 '4': balance728 '5': business_loan729 '6': card_issues730 '7': cash_deposit731 '8': direct_debit732 '9': freeze733 '10': high_value_payment734 '11': joint_account735 '12': latest_transactions736 '13': pay_bill737 - name: lang_id738 dtype:739 class_label:740 names:741 '0': cs-CZ742 '1': de-DE743 '2': en-AU744 '3': en-GB745 '4': en-US746 '5': es-ES747 '6': fr-FR748 '7': it-IT749 '8': ko-KR750 '9': nl-NL751 '10': pl-PL752 '11': pt-PT753 '12': ru-RU754 '13': zh-CN755 splits:756 - name: train757 num_bytes: 88547861.0758 num_examples: 562759 download_size: 52662883760 dataset_size: 88547861.0761- config_name: pt-PT762 features:763 - name: path764 dtype: string765 - name: audio766 dtype:767 audio:768 sampling_rate: 8000769 - name: transcription770 dtype: string771 - name: english_transcription772 dtype: string773 - name: intent_class774 dtype:775 class_label:776 names:777 '0': abroad778 '1': address779 '2': app_error780 '3': atm_limit781 '4': balance782 '5': business_loan783 '6': card_issues784 '7': cash_deposit785 '8': direct_debit786 '9': freeze787 '10': high_value_payment788 '11': joint_account789 '12': latest_transactions790 '13': pay_bill791 - name: lang_id792 dtype:793 class_label:794 names:795 '0': cs-CZ796 '1': de-DE797 '2': en-AU798 '3': en-GB799 '4': en-US800 '5': es-ES801 '6': fr-FR802 '7': it-IT803 '8': ko-KR804 '9': nl-NL805 '10': pl-PL806 '11': pt-PT807 '12': ru-RU808 '13': zh-CN809 splits:810 - name: train811 num_bytes: 77549349.0812 num_examples: 604813 download_size: 51738310814 dataset_size: 77549349.0815- config_name: ru-RU816 features:817 - name: path818 dtype: string819 - name: audio820 dtype:821 audio:822 sampling_rate: 8000823 - name: transcription824 dtype: string825 - name: english_transcription826 dtype: string827 - name: intent_class828 dtype:829 class_label:830 names:831 '0': abroad832 '1': address833 '2': app_error834 '3': atm_limit835 '4': balance836 '5': business_loan837 '6': card_issues838 '7': cash_deposit839 '8': direct_debit840 '9': freeze841 '10': high_value_payment842 '11': joint_account843 '12': latest_transactions844 '13': pay_bill845 - name: lang_id846 dtype:847 class_label:848 names:849 '0': cs-CZ850 '1': de-DE851 '2': en-AU852 '3': en-GB853 '4': en-US854 '5': es-ES855 '6': fr-FR856 '7': it-IT857 '8': ko-KR858 '9': nl-NL859 '10': pl-PL860 '11': pt-PT861 '12': ru-RU862 '13': zh-CN863 splits:864 - name: train865 num_bytes: 38159953.0866 num_examples: 539867 download_size: 34494545868 dataset_size: 38159953.0869- config_name: zh-CN870 features:871 - name: path872 dtype: string873 - name: audio874 dtype:875 audio:876 sampling_rate: 8000877 - name: transcription878 dtype: string879 - name: english_transcription880 dtype: string881 - name: intent_class882 dtype:883 class_label:884 names:885 '0': abroad886 '1': address887 '2': app_error888 '3': atm_limit889 '4': balance890 '5': business_loan891 '6': card_issues892 '7': cash_deposit893 '8': direct_debit894 '9': freeze895 '10': high_value_payment896 '11': joint_account897 '12': latest_transactions898 '13': pay_bill899 - name: lang_id900 dtype:901 class_label:902 names:903 '0': cs-CZ904 '1': de-DE905 '2': en-AU906 '3': en-GB907 '4': en-US908 '5': es-ES909 '6': fr-FR910 '7': it-IT911 '8': ko-KR912 '9': nl-NL913 '10': pl-PL914 '11': pt-PT915 '12': ru-RU916 '13': zh-CN917 splits:918 - name: train919 num_bytes: 36522343.0920 num_examples: 502921 download_size: 32357378922 dataset_size: 36522343.0923---924 925# MInDS-14926 927## Dataset Description928 929- **Fine-Tuning script:** [pytorch/audio-classification](https://github.com/huggingface/transformers/tree/main/examples/pytorch/audio-classification)930- **Paper:** [Multilingual and Cross-Lingual Intent Detection from Spoken Data](https://arxiv.org/abs/2104.08524)931- **Total amount of disk used:** ca. 500 MB 932 933MINDS-14 is training and evaluation resource for intent detection task with spoken data. It covers 14 934intents extracted from a commercial system in the e-banking domain, associated with spoken examples in 14 diverse language varieties.935 936## Example937 938MInDS-14 can be downloaded and used as follows:939 940```py941from datasets import load_dataset942 943minds_14 = load_dataset("PolyAI/minds14", "fr-FR") # for French944# to download all data for multi-lingual fine-tuning uncomment following line945# minds_14 = load_dataset("PolyAI/all", "all")946 947# see structure948print(minds_14)949 950# load audio sample on the fly951audio_input = minds_14["train"][0]["audio"] # first decoded audio sample952intent_class = minds_14["train"][0]["intent_class"] # first transcription953intent = minds_14["train"].features["intent_class"].names[intent_class]954 955# use audio_input and language_class to fine-tune your model for audio classification956```957 958## Dataset Structure959 960We show detailed information the example configurations `fr-FR` of the dataset.961All other configurations have the same structure.962 963### Data Instances964 965**fr-FR**966 967- Size of downloaded dataset files: 471 MB968- Size of the generated dataset: 300 KB969- Total amount of disk used: 471 MB970 971 972An example of a datainstance of the config `fr-FR` looks as follows:973 974```975{976 "path": "fr-FR~ADDRESS/response_4.wav",977 "audio": {978 "array": array(979 [0.0, 0.0, 0.0, ..., 0.0, 0.00048828, -0.00024414], dtype=float32980 ),981 "sampling_rate": 8000,982 },983 "transcription": "je souhaite changer mon adresse",984 "english_transcription": "I want to change my address",985 "intent_class": 1,986 "lang_id": 6,987}988```989 990### Data Fields991The data fields are the same among all splits.992 993- **path** (str): Path to the audio file994- **audio** (dict): Audio object including loaded audio array, sampling rate and path ot audio995- **transcription** (str): Transcription of the audio file996- **english_transcription** (str): English transcription of the audio file997- **intent_class** (int): Class id of intent998- **lang_id** (int): Id of language999 1000### Data Splits1001Every config only has the `"train"` split containing of *ca.* 600 examples.1002 1003## Dataset Creation1004 1005[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)1006 1007## Considerations for Using the Data1008 1009### Social Impact of Dataset1010 1011[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)1012 1013### Discussion of Biases1014 1015[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)1016 1017### Other Known Limitations1018 1019[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)1020 1021## Additional Information1022 1023### Dataset Curators1024 1025[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)1026 1027### Licensing Information1028 1029All datasets are licensed under the [Creative Commons license (CC-BY)](https://creativecommons.org/licenses/).1030 1031### Citation Information1032 1033```1034@article{DBLP:journals/corr/abs-2104-08524,1035 author = {Daniela Gerz and1036 Pei{-}Hao Su and1037 Razvan Kusztos and1038 Avishek Mondal and1039 Michal Lis and1040 Eshan Singhal and1041 Nikola Mrksic and1042 Tsung{-}Hsien Wen and1043 Ivan Vulic},1044 title = {Multilingual and Cross-Lingual Intent Detection from Spoken Data},1045 journal = {CoRR},1046 volume = {abs/2104.08524},1047 year = {2021},1048 url = {https://arxiv.org/abs/2104.08524},1049 eprinttype = {arXiv},1050 eprint = {2104.08524},1051 timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},1052 biburl = {https://dblp.org/rec/journals/corr/abs-2104-08524.bib},1053 bibsource = {dblp computer science bibliography, https://dblp.org}1054}1055```1056 1057### Contributions1058 1059Thanks to [@patrickvonplaten](https://github.com/patrickvonplaten) for adding this dataset1060