{"url":"/dataset/must-c","name":"MuST-C","full_name":null,"description_markdown":"**MuST-C** currently represents the largest publicly available multilingual corpus (one-to-many) for speech translation. It covers eight language directions, from English to German, Spanish, French, Italian, Dutch, Portuguese, Romanian and Russian. The corpus consists of audio, transcriptions and translations of English TED talks, and it comes with a predefined training, validation and test split.\r\n\r\nSource: [One-to-Many Multilingual End-to-End Speech Translation](https://arxiv.org/abs/1910.03320)\r\nImage Source: [https://mt.fbk.eu/must-c](https://mt.fbk.eu/must-c)","description_withheld":null,"homepage":"https://mt.fbk.eu/must-c","introduced_date":"2019-01-01","introduced_date_note":null,"introduced_by":{"paper":"/paper/must-c-a-multilingual-speech-translation","title":"MuST-C: a Multilingual Speech Translation Corpus","first_author":"Mattia A. Di Gangi","url":null},"license":{"name":"CC BY-NC-ND 4.0","url":"https://creativecommons.org/licenses/by-nc-nd/4.0/"},"modalities":[{"name":"Texts","url":"/datasets/modality/texts"},{"name":"Audio","url":"/datasets/modality/audio"}],"tasks":[{"name":"Speech Recognition","url":"/task/speech-recognition","datasets_with_task":"/datasets/task/speech-recognition"},{"name":"Speech-to-Text Translation","url":"/task/speech-to-text-translation","datasets_with_task":"/datasets/task/speech-to-text-translation"},{"name":"Data Augmentation","url":"/task/data-augmentation","datasets_with_task":"/datasets/task/data-augmentation"}],"languages":[{"name":"French","url":"/datasets/language/french"},{"name":"Spanish","url":"/datasets/language/spanish"},{"name":"German","url":"/datasets/language/german"},{"name":"Italian","url":"/datasets/language/italian"},{"name":"Chinese","url":"/datasets/language/chinese"},{"name":"Russian","url":"/datasets/language/russian"},{"name":"Portuguese","url":"/datasets/language/portuguese"},{"name":"Arabic","url":"/datasets/language/arabic"},{"name":"Czech","url":"/datasets/language/czech"},{"name":"Dutch","url":"/datasets/language/dutch"},{"name":"Persian","url":"/datasets/language/persian"},{"name":"Romanian","url":"/datasets/language/romanian"},{"name":"Turkish","url":"/datasets/language/turkish"},{"name":"Vietnamese","url":"/datasets/language/vietnamese"}],"variants":["MuST-C EN->DE","MuST-C"],"data_loaders":[{"repo":"https://github.com/RXrzh/MuST-C_Data","url":"https://github.com/RXrzh/MuST-C_Data","frameworks":["pytorch"]}],"num_papers_in_archive":216,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/speech-to-text-translation-on-must-c-en-de","task":"Speech-to-Text Translation","dataset_variant":"MuST-C EN->DE","rows":8,"metrics":["Case-sensitive sacreBLEU"],"first_row_in_archive_order":{"model":"Task Modulation + Multitask Learning(ASR/MT) + Data Augmentation","paper":"/paper/task-aware-multi-task-learning-for-speech-to","metrics":{"Case-sensitive sacreBLEU":"28.88"},"code_links":[]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/speech-to-text-translation-on-must-c-1","task":"Speech-to-Text Translation","dataset_variant":"MuST-C","rows":2,"metrics":["SacreBLEU"],"first_row_in_archive_order":{"model":"Transformer with Adapters","paper":"/paper/lightweight-adapter-tuning-for-multilingual","metrics":{"SacreBLEU":"26.61"},"code_links":[{"title":"formiel/fairseq","url":"https://github.com/formiel/fairseq"},{"title":"formiel/fairseq","url":"https://github.com/formiel/fairseq/blob/master/examples/speech_to_text/docs/adapters.md"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/speechformer-reducing-information-loss-in","title":"Speechformer: Reducing Information Loss in Direct Speech Translation","date":"2021-09-09","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":5,"samples_ran":2,"samples_unverified":3,"pointer_only_for_licence":4,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/task-aware-multi-task-learning-for-speech-to","title":"TASK AWARE MULTI-TASK LEARNING FOR SPEECH TO TEXT TASKS","date":"2021-06-10","rows_on_this_dataset":1,"code_links":0,"syntology":null},{"paper":"/paper/lightweight-adapter-tuning-for-multilingual","title":"Lightweight Adapter Tuning for Multilingual Speech Translation","date":"2021-06-02","rows_on_this_dataset":2,"code_links":2,"syntology":null},{"paper":"/paper/upc-s-speech-translation-system-for-iwslt","title":"End-to-End Speech Translation with Pre-trained Models and Adapters: UPC at IWSLT 2021","date":"2021-05-10","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/neurst-neural-speech-translation-toolkit","title":"NeurST: Neural Speech Translation Toolkit","date":"2020-12-18","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/dual-decoder-transformer-for-joint-automatic","title":"Dual-decoder Transformer for Joint Automatic Speech Recognition and Multilingual Speech Translation","date":"2020-11-02","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/fairseq-s2t-fast-speech-to-text-modeling-with","title":"fairseq S2T: Fast Speech-to-Text Modeling with fairseq","date":"2020-10-11","rows_on_this_dataset":1,"code_links":5,"syntology":null},{"paper":"/paper/end-to-end-offline-speech-translation-system","title":"End-to-End Offline Speech Translation System for IWSLT 2020 using Modality Agnostic Meta-Learning","date":"2020-07-01","rows_on_this_dataset":1,"code_links":0,"syntology":null}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":1,"samples_harvested":5,"samples_ran":2,"samples_unverified":3,"pointer_only_for_licence":4,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}