{"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/a-segmental-framework-for-fully-unsupervised","title":"A segmental framework for fully-unsupervised large-vocabulary speech recognition","arxiv_id":"1606.06950","date":"2016-06-22","proceeding":null,"authors":["Herman Kamper","Aren Jansen","Sharon Goldwater"],"abstract":"Zero-resource speech technology is a growing research area that aims to\ndevelop methods for speech processing in the absence of transcriptions,\nlexicons, or language modelling text. Early term discovery systems focused on\nidentifying isolated recurring patterns in a corpus, while more recent\nfull-coverage systems attempt to completely segment and cluster the audio into\nword-like units---effectively performing unsupervised speech recognition. This\narticle presents the first attempt we are aware of to apply such a system to\nlarge-vocabulary multi-speaker data. Our system uses a Bayesian modelling\nframework with segmental word representations: each word segment is represented\nas a fixed-dimensional acoustic embedding obtained by mapping the sequence of\nfeature frames to a single embedding vector. We compare our system on English\nand Xitsonga datasets to state-of-the-art baselines, using a variety of\nmeasures including word error rate (obtained by mapping the unsupervised output\nto ground truth transcriptions). Very high word error rates are reported---in\nthe order of 70--80% for speaker-dependent and 80--95% for speaker-independent\nsystems---highlighting the difficulty of this task. Nevertheless, in terms of\ncluster quality and word segmentation metrics, we show that by imposing a\nconsistent top-down segmentation while also using bottom-up knowledge from\ndetected syllable boundaries, both single-speaker and multi-speaker versions of\nour system outperform a purely bottom-up single-speaker syllable-based\napproach. We also show that the discovered clusters can be made less speaker-\nand gender-specific by using an unsupervised autoencoder-like feature extractor\nto learn better frame-level features (prior to embedding). Our system's\ndiscovered clusters are still less pure than those of unsupervised term\ndiscovery systems, but provide far greater coverage.","url_abs":"http://arxiv.org/abs/1606.06950v2","url_pdf":"http://arxiv.org/pdf/1606.06950v2.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":"a-segmental-framework-for-fully-unsupervised","repo_url":"https://github.com/kamperh/bucktsong_segmentalist","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"a-segmental-framework-for-fully-unsupervised","repo_url":"https://github.com/kamperh/bucktsong_eskmeans","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"a-segmental-framework-for-fully-unsupervised","repo_url":"https://github.com/kamperh/recipe_bucktsong_awe","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"a-segmental-framework-for-fully-unsupervised","repo_url":"https://github.com/kamperh/recipe_bucktsong_awe_py3","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"a-segmental-framework-for-fully-unsupervised","repo_url":"https://github.com/quantum-fusion/AI-deeplearning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"speech-recognition","task_name":"Speech Recognition"},{"task_slug":"unsupervised-speech-recognition","task_name":"Unsupervised Speech Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1606.06950","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1606.06950"}},"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/kamperh/bucktsong_eskmeans","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kamperh/recipe_bucktsong_awe_py3","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kamperh/recipe_bucktsong_awe","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kamperh/bucktsong_segmentalist","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/quantum-fusion/AI-deeplearning","reach":{"status":"ok"}}],"summary":{"ran_draft_wrong":1,"ran_honours":1},"by_repo_kind":{"listed":{"samples":2,"ran":2,"repositories":1}},"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":2,"samples":[{"code_sha256_prefix":"5c3b6373d392daf2","entry":"buckeye_utt_to_label","repo":"kamperh/recipe_bucktsong_awe_py3","repo_kind":"listed","path":"downsample/analyse_embeds.py","file_url":"https://github.com/kamperh/recipe_bucktsong_awe_py3/blob/HEAD/downsample/analyse_embeds.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"5c3b6373d392daf2"}},{"code_sha256_prefix":"9eff407372d4ffbd","entry":"get_embeds_and_labels","repo":"kamperh/recipe_bucktsong_awe_py3","repo_kind":"listed","path":"downsample/analyse_embeds.py","file_url":"https://github.com/kamperh/recipe_bucktsong_awe_py3/blob/HEAD/downsample/analyse_embeds.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"9eff407372d4ffbd"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}