{"success":true,"data":{"author":"Mark Mazumder, Sharad Chitlangia, Colby Banbury et al. (MLCommons, Harvard)","bucketAccessType":"public","bucketName":"MLCommons/ml_spoken_words","bucketPrefix":"","category":"Speech & Audio","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:17.650Z","description":"The Multilingual Spoken Words Corpus is a large audio dataset of spoken words in 50 languages, built for keyword spotting and spoken term search. It contains more than 340,000 keywords and about 23.4 million one-second spoken examples, over 6,000 hours in total. The examples were generated by applying forced alignment to crowd-sourced sentence-level recordings from Mozilla Common Voice to obtain per-word timings and then extracting each word; all alignments are included. Use cases include on-device wake-word and keyword-spotting models, few-shot keyword spotting in low-resource languages and spoken term detection. This repository stores the audio in two formats, 16 kHz wav and 48 kHz opus, under data/wav and data/opus with one folder per language (50 each), and train/dev/test split files under data/splits (measured 2026-11-14; the Hub reports about 930 GB). Configurations are named '{lang}_{format}', for example datasets.load_dataset('MLCommons/ml_spoken_words', 'tt_wav'). Published by MLCommons under CC BY 4.0 for academic research and commercial applications; cite Mazumder et al., NeurIPS 2021 Datasets and Benchmarks.","displayTitle":"Multilingual Spoken Words Corpus (MSWC)","id":"6dcf302f-db6e-539b-bd80-89e8cd63c705","license":"CC-BY-4.0","licenseEvidenceUrl":"https://huggingface.co/datasets/MLCommons/ml_spoken_words","objectCountEstimate":1333,"probeStatus":"ok","referenceUrl":"https://datasets-benchmarks-proceedings.neurips.cc/paper/2021/file/fe131d7f5a6b38b23cc967316c13dae2-Paper-round2.pdf","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"MLCommons/ml_spoken_words\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"MLCommons/ml_spoken_words\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=MLCommons, license:cc-by-4.0, size_categories:10M<n<100M, downloads=9894"],"format":"mixed","generatedAt":"2026-11-14T14:15:30.467Z","layout":{"example":"data/opus/ar/dev/audio/0.tar.gz","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/opus/ar/dev/audio/0.tar.gz","truncatedAt":1536}],"sampleObjects":["data/.gitattributes","data/opus/ar/dev/audio/0.tar.gz","data/opus/ar/dev/n_files.txt","data/opus/ar/test/audio/0.tar.gz","data/opus/ar/test/n_files.txt"],"sidecars":[{"bytes":"---\nannotations_creators:\n- machine-generated\nlanguage_creators:\n- other\nlanguage:\n- ar\n- as\n- br\n- ca\n- cnh\n- cs\n- cv\n- cy\n- de\n- dv\n- el\n- en\n- eo\n- es\n- et\n- eu\n- fa\n- fr\n- fy\n- ga\n- gn\n- ha\n- ia\n- id\n- it\n- ka\n- ky\n- lt\n- lv\n- mn\n- mt\n- nl\n- or\n- pl\n- pt\n- rm\n- ro\n- ru\n- rw\n- sah\n- sk\n- sl\n- sv\n- ta\n- tr\n- tt\n- uk\n- vi\n- zh\nlicense:\n- cc-by-4.0\nmultilinguality:\n- multilingual\nsize_categories:\n- 10M<n<100M\nsource_datasets:\n- extended|common_voice\ntask_categories:\n- audio-classification\ntask_ids: []\npretty_name: Multilingual Spoken Words\nlanguage_bcp47:\n- fy-NL\n- ga-IE\n- rm-sursilv\n- rm-vallader\n- sv-SE\n- zh-CN\ntags:\n- other-keyword-spotting\n---\n\n# Dataset Card for Multilingual Spoken Words\n\n## Table of Contents\n- [Table of Contents](#table-of-contents)\n- [Dataset Description](#dataset-description)\n  - [Dataset Summary](#dataset-summary)\n  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)\n  - [Languages](#languages)\n- [Dataset Structure](#dataset-structure)\n  - [Data Instances](#data-instances)\n  - [Data Fields](#data-fields)\n  - [Data Splits](#data-splits)\n- [Dataset Creation](#dataset-creation)\n  - [Curation Rationale](#curation-rationale)\n  - [Source Data](#source-data)\n  - [Annotations](#annotations)\n  - [Personal and Sensitive Information](#personal-and-sensitive-information)\n- [Considerations for Using the Data](#considerations-for-using-the-data)\n  - [Social Impact of Dataset](#social-impact-of-dataset)\n  - [Discussion of Biases](#discussion-of-biases)\n  - [Other Known Limitations](#other-known-limitations)\n- [Additional Information](#additional-information)\n  - [Dataset Curators](#dataset-curators)\n  - [Licensing Information](#licensing-information)\n  - [Citation Information](#citation-information)\n  - [Contributions](#contributions)\n\n## Dataset Description\n\n- **Homepage:** https://mlcommons.org/en/multilingual-spoken-words/\n- **Repository:** https://github.com/harvard-edge/multilingual_kws\n- **Paper:** https://datasets-benchmarks-proceedings.neurips.cc/paper/2021/file/fe131d7f5a6b38b23cc967316c13dae2-Paper-round2.pdf\n- **Leaderboard:**\n- **Point of Contact:**\n\n### Dataset Summary\n\nMultilingual Spoken Words Corpus is a large and growing audio dataset of spoken\nwords in 50 languages collectively spoken by over 5 billion people, for academic\nresearch and commercial applications in keyword spotting and spoken term search,\nlicensed under CC-BY 4.0. The dataset contains more than 340,000 keywords,\ntotaling 23.4 million 1-second spoken examples (over 6,000 hours). The dataset\nhas many use cases, ranging from voice-enabled consumer devices to call center\nautomation. This dataset is generated by applying forced alignment on crowd-sourced sentence-level \naudio to produce per-word timing estimates for extraction.\nAll alignments are included in the dataset.\n\nData is provided in two formats: `wav` (16KHz) and `opus` (48KHz). Default configurations look like \n`\"{lang}_{format}\"`, so to load, for example, Tatar in wav format do:\n\n```python\nds = load_dataset(\"MLCommons/ml_spoken_words\", \"tt_wav\")\n```\n\nTo download multiple languages in a single dataset pass list of languages to `languages` argument:\n```python\nds = load_dataset(\"MLCommons/ml_spoken_words\", languages=[\"ar\", \"tt\", \"br\"])\n```\n\nTo download a specific format pass it to the `format` argument (default format is `wav`):\n```python\nds = load_dataset(\"MLCommons/ml_spoken_words\", languages=[\"ar\", \"tt\", \"br\"], format=\"opus\")\n```\nNote that each time you provide different sets of languages, \nexamples are generated from scratch even if you already provided one or several of them before \nbecause custom configurations are created each time (the data is **not** redownloaded though).\n\n### Supported Tasks and Leaderboards\n\nKeyword spotting, Spoken term search\n\n### Languages\n\nThe dataset is multilingual. To specify several languages to download pass a list of them to the\n`languages` argument:\n\n```python\nds = load_dataset(\"MLCommons/ml_spoken_words\", languages=[\"ar\", \"tt\", \"br\"])\n```\n\nThe dataset contains data for the following languages:\n\nLow-resourced (<10 hou","encoding":"utf8","key":"README.md","name":"README.md","size":8612,"truncated":true}],"sourceObjectKey":"data/opus/ar/dev/audio/0.tar.gz","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"MLCommons/ml_spoken_words\", filename=\"data/opus/ar/dev/audio/0.tar.gz\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"data/opus/ar/dev/n_files.txt","totalObjects":1333},"summary":"The Multilingual Spoken Words Corpus is a large audio dataset of spoken words in 50 languages, built for keyword spotting and spoken term search.","title":"Multilingual Spoken Words Corpus (MSWC)","updatedAt":"2026-11-14T17:39:43.674Z","visibility":"public","websiteUrl":"https://mlcommons.org/en/multilingual-spoken-words/"}}