{"url":"/dataset/voxceleb2","name":"VoxCeleb2","full_name":"VoxCeleb2","description_markdown":"**VoxCeleb2** is a large scale speaker recognition dataset obtained automatically from open-source media. VoxCeleb2 consists of over a million utterances from over 6k speakers. Since the dataset is collected ‘in the wild’, the speech segments are corrupted with real world noise including laughter, cross-talk, channel effects, music and other sounds. The dataset is also multilingual, with speech from speakers of 145 different nationalities, covering a wide range of accents, ages, ethnicities and languages. The dataset is audio-visual, so is also useful for a number of other applications, for example – visual speech synthesis, speech separation, cross-modal transfer from face to voice or vice versa and training face recognition from video to complement existing face recognition datasets.\r\n\r\nSource: [VoxCeleb2: Deep Speaker Recognition](https://arxiv.org/pdf/1806.05622v2.pdf)\r\nImage Source: [https://www.robots.ox.ac.uk/~vgg/data/voxceleb/](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/)","description_withheld":null,"homepage":"https://www.robots.ox.ac.uk/~vgg/data/voxceleb/","introduced_date":"2018-01-01","introduced_date_note":null,"introduced_by":{"paper":"/paper/voxceleb2-deep-speaker-recognition","title":"VoxCeleb2: Deep Speaker Recognition","first_author":"Joon Son Chung","url":null},"license":{"name":"CC BY 4.0","url":"https://www.robots.ox.ac.uk/~vgg/data/voxceleb/files/license.txt"},"modalities":[{"name":"Images","url":"/datasets/modality/images"},{"name":"Videos","url":"/datasets/modality/videos"},{"name":"Texts","url":"/datasets/modality/texts"},{"name":"Audio","url":"/datasets/modality/audio"}],"tasks":[{"name":"Speech Separation","url":"/task/speech-separation","datasets_with_task":"/datasets/task/speech-separation"},{"name":"Speaker Verification","url":"/task/speaker-verification","datasets_with_task":"/datasets/task/speaker-verification"},{"name":"Talking Head Generation","url":"/task/talking-head-generation","datasets_with_task":"/datasets/task/talking-head-generation"}],"languages":[{"name":"Multilingual","url":"/datasets/language/multilingual"}],"variants":["VoxCeleb2 - 1-shot learning","VoxCeleb2 - 8-shot learning","VoxCeleb2 - 32-shot learning","VoxCeleb2"],"data_loaders":[],"num_papers_in_archive":564,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/speech-separation-on-voxceleb2","task":"Speech Separation","dataset_variant":"VoxCeleb2","rows":5,"metrics":["SI-SNRi","SDRi"],"first_row_in_archive_order":{"model":"IIANet","paper":"/paper/scanet-a-self-and-cross-attention-network-for","metrics":{"SDRi":"15.1","SI-SNRi":"14.0"},"code_links":[{"title":"JusperLee/IIANet","url":"https://github.com/JusperLee/IIANet"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/talking-head-generation-on-voxceleb2-1-shot","task":"Talking Head Generation","dataset_variant":"VoxCeleb2 - 1-shot learning","rows":5,"metrics":["CSIM","LPIPS","Normalized Pose Error","SSIM","inference time (ms)","FID"],"first_row_in_archive_order":{"model":"Fast Bi-layer Avatars (medium size)","paper":"/paper/fast-bi-layer-neural-synthesis-of-one-shot","metrics":{"CSIM":"0.653","LPIPS":"0.358","Normalized Pose Error":"43.3","SSIM":"0.508","inference time (ms)":"4"},"code_links":[{"title":"saic-violet/bilayer-model","url":"https://github.com/saic-violet/bilayer-model"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/talking-head-generation-on-voxceleb2-8-shot","task":"Talking Head Generation","dataset_variant":"VoxCeleb2 - 8-shot learning","rows":2,"metrics":["FID"],"first_row_in_archive_order":{"model":"CainGAN","paper":"/paper/pose-manipulation-with-identity-preservation-1","metrics":{"FID":"24.9"},"code_links":[]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/speaker-verification-on-voxceleb2","task":"Speaker Verification","dataset_variant":"VoxCeleb2","rows":1,"metrics":["EER"],"first_row_in_archive_order":{"model":"ResNet-50","paper":"/paper/voxceleb2-deep-speaker-recognition","metrics":{"EER":"100"},"code_links":[{"title":"a-nagrani/VGGVox","url":"https://github.com/a-nagrani/VGGVox"},{"title":"MainRo/deep-speaker","url":"https://github.com/MainRo/deep-speaker"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/talking-head-generation-on-voxceleb2-32-shot","task":"Talking Head Generation","dataset_variant":"VoxCeleb2 - 32-shot learning","rows":1,"metrics":["FID"],"first_row_in_archive_order":{"model":"Few-shot Adversarial Model","paper":"/paper/few-shot-adversarial-learning-of-realistic","metrics":{"FID":"30.6"},"code_links":[{"title":"vincent-thevenin/Realistic-Neural-Talking-Head-Models","url":"https://github.com/vincent-thevenin/Realistic-Neural-Talking-Head-Models"},{"title":"grey-eye/talking-heads","url":"https://github.com/grey-eye/talking-heads"},{"title":"shoutOutYangJie/Few-Shot-Adversarial-Learning-for-face-swap","url":"https://github.com/shoutOutYangJie/Few-Shot-Adversarial-Learning-for-face-swap"},{"title":"ZVK/Talking-Heads","url":"https://github.com/ZVK/Talking-Heads"},{"title":"ZVK/talking_heads","url":"https://github.com/ZVK/talking_heads"},{"title":"Ierezell/PapierFewShot","url":"https://github.com/Ierezell/PapierFewShot"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/rtfs-net-recurrent-time-frequency-modelling","title":"RTFS-Net: Recurrent Time-Frequency Modelling for Efficient Audio-Visual Speech Separation","date":"2023-09-29","rows_on_this_dataset":3,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":1,"samples_ran":0,"samples_unverified":1,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/scanet-a-self-and-cross-attention-network-for","title":"IIANet: An Intra- and Inter-Modality Attention Network for Audio-Visual Speech Separation","date":"2023-08-16","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":1,"samples_ran":1,"samples_unverified":0,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/an-audio-visual-speech-separation-model","title":"An Audio-Visual Speech Separation Model Inspired by Cortico-Thalamo-Cortical Circuits","date":"2022-12-21","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":2,"samples_unverified":1,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/fast-bi-layer-neural-synthesis-of-one-shot","title":"Fast Bi-layer Neural Synthesis of One-Shot Realistic Head Avatars","date":"2020-08-24","rows_on_this_dataset":3,"code_links":1,"syntology":null},{"paper":"/paper/pose-manipulation-with-identity-preservation-1","title":"Pose Manipulation with Identity Preservation","date":"2020-04-20","rows_on_this_dataset":2,"code_links":0,"syntology":null},{"paper":"/paper/few-shot-adversarial-learning-of-realistic","title":"Few-Shot Adversarial Learning of Realistic Neural Talking Head Models","date":"2019-05-20","rows_on_this_dataset":3,"code_links":6,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":2,"samples_ran":2,"samples_unverified":0,"pointer_only_for_licence":2,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/voxceleb2-deep-speaker-recognition","title":"VoxCeleb2: Deep Speaker Recognition","date":"2018-06-14","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":13,"samples_ran":0,"samples_unverified":13,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":5,"samples_harvested":20,"samples_ran":5,"samples_unverified":15,"pointer_only_for_licence":2,"papers_with_no_sample_that_ran":2,"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."}