{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/multi-talker-speech-separation-with-utterance","title":"Multi-talker Speech Separation with Utterance-level Permutation Invariant Training of Deep Recurrent Neural Networks","arxiv_id":"1703.06284","date":"2017-03-18","proceeding":null,"authors":["Morten Kolbæk","Dong Yu","Zheng-Hua Tan","Jesper Jensen"],"abstract":"In this paper we propose the utterance-level Permutation Invariant Training\n(uPIT) technique. uPIT is a practically applicable, end-to-end, deep learning\nbased solution for speaker independent multi-talker speech separation.\nSpecifically, uPIT extends the recently proposed Permutation Invariant Training\n(PIT) technique with an utterance-level cost function, hence eliminating the\nneed for solving an additional permutation problem during inference, which is\notherwise required by frame-level PIT. We achieve this using Recurrent Neural\nNetworks (RNNs) that, during training, minimize the utterance-level separation\nerror, hence forcing separated frames belonging to the same speaker to be\naligned to the same output stream. In practice, this allows RNNs, trained with\nuPIT, to separate multi-talker mixed speech without any prior knowledge of\nsignal duration, number of speakers, speaker identity or gender. We evaluated\nuPIT on the WSJ0 and Danish two- and three-talker mixed-speech separation tasks\nand found that uPIT outperforms techniques based on Non-negative Matrix\nFactorization (NMF) and Computational Auditory Scene Analysis (CASA), and\ncompares favorably with Deep Clustering (DPCL) and the Deep Attractor Network\n(DANet). Furthermore, we found that models trained with uPIT generalize well to\nunseen speakers and languages. Finally, we found that a single model, trained\nwith uPIT, can handle both two-speaker, and three-speaker speech mixtures.","url_abs":"http://arxiv.org/abs/1703.06284v2","url_pdf":"http://arxiv.org/pdf/1703.06284v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"multi-talker-speech-separation-with-utterance","repo_url":"https://github.com/fchest/uPIT","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"multi-talker-speech-separation-with-utterance","repo_url":"https://github.com/snsun/pit-speech-separation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"multi-talker-speech-separation-with-utterance","repo_url":"https://github.com/JusperLee/UtterancePIT-Speech-Separation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"deep-clustering","task_name":"Deep Clustering"},{"task_slug":"speech-separation","task_name":"Speech Separation"}],"methods":[{"method_slug":"upit","method_name":"uPIT"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.06284","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1703.06284"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/fchest/uPIT","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/JusperLee/UtterancePIT-Speech-Separation","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/snsun/pit-speech-separation","reach":{"status":"ok"}}],"summary":{"ran_draft_wrong":1,"unverified":2},"by_repo_kind":{"listed":{"samples":3,"ran":1,"repositories":2}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":3,"samples":[{"code_sha256_prefix":"e88f99dab9684237","entry":"packed_sequence_cuda","repo":"JusperLee/UtterancePIT-Speech-Separation","repo_kind":"listed","path":"trainer.py","file_url":"https://github.com/JusperLee/UtterancePIT-Speech-Separation/blob/HEAD/trainer.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"e88f99dab9684237"}},{"code_sha256_prefix":"d91435f239876721","entry":"eval_one_epoch","repo":"fchest/uPIT","repo_kind":"listed","path":"run_lstm.py","file_url":"https://github.com/fchest/uPIT/blob/HEAD/run_lstm.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"d91435f239876721"}},{"code_sha256_prefix":"c5b2577f92fc75b7","entry":"train_one_epoch","repo":"fchest/uPIT","repo_kind":"listed","path":"run_lstm.py","file_url":"https://github.com/fchest/uPIT/blob/HEAD/run_lstm.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"c5b2577f92fc75b7"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}