{"corpus_id":54763794,"paper_sha":"eabbd27e3098148d8e14202122f6a3109ee7f1ec","doi":"10.15598/AEEE.V9I5.545","arxiv_id":null,"pmid":null,"pmcid":null,"mag_id":2073565062,"dblp_id":null,"acl_id":null,"title":"Comparison of Current Frame-Based Phoneme Classifiers","year":2011,"publication_date":"2011-12-24","venue":"","journal":{"name":"Advances in Electrical and Electronic Engineering","pages":"243-250","volume":"9"},"journal_issn":null,"journal_title":null,"publication_types":[],"pubmed_pub_types":null,"s2_fields_of_study":["Computer Science"],"reference_count":21,"citation_count":2,"influential_citation_count":0,"is_open_access":false,"arxiv_categories":null,"arxiv_license":null,"arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":null,"s2_open_access_landing_url":null,"s2_open_access_license":null,"s2_open_access_status":null,"pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"This paper discusses current approaches for frame-based classification and evaluates today's most common frame-based classifiers. These classifiers can be divided into the two main groups - generic classifiers which create the most probable model based on the training data (for example GMM) and discriminative classifiers which focus on creating decision hyper plane (SVM based methods). A lot of research has been done with the generic classifiers and therefore this paper will be mainly focused on the discriminative classifiers. Four discriminative classifiers are presented - two linear and two non-linear. All of these discriminative classifiers implement a hierarchical tree root structure over the input phoneme group which shown to be an effective. Moreover, two efficient training algorithms are presented. First, we demonstrate advantages of discriminative classifiers by comparison with a standard generic classifier represented by a GMM. Second, we show benefits of our proposed training algorithm. All tests were performed for English only - over the TIMIT speech corpus (corpus of Native American speakers).","claims":[{"public_id":"cl_342d0ea83d0ed57db4543218e82671ff","status":"active","text":"Discriminative classifiers show advantages over a standard generic classifier represented by a GMM.","confidence":0.89,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_342d0ea83d0ed57db4543218e82671ff"},{"public_id":"cl_4614aacb7be2ad46e228288f72fe6805","status":"active","text":"Four discriminative frame-based phoneme classifiers are presented, including two linear and two non-linear variants.","confidence":0.97,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_4614aacb7be2ad46e228288f72fe6805"},{"public_id":"cl_cc24498e93ab26b6dbe6e2796bcbbc03","status":"active","text":"The discriminative classifiers use a hierarchical tree-root structure over the input phoneme group, and this structure is reported as effective.","confidence":0.92,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_cc24498e93ab26b6dbe6e2796bcbbc03"},{"public_id":"cl_c6d955b81168d7d572dba1588ee9aa7a","status":"active","text":"The proposed training algorithm provides benefits in the reported tests on English speech from the TIMIT corpus.","confidence":0.9,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_c6d955b81168d7d572dba1588ee9aa7a"},{"public_id":"cl_a8608c90b5757012780a79e338f140f4","status":"active","text":"Two efficient training algorithms are introduced for the proposed discriminative classifiers.","confidence":0.96,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_a8608c90b5757012780a79e338f140f4"}],"concepts":[{"public_id":"co_01cc92e49101785ae47c48d7fd83fa0e","status":"active","name":"generic classifiers","description":"Classification methods that model class distributions from training data to estimate the most probable model.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_01cc92e49101785ae47c48d7fd83fa0e"},{"public_id":"co_20ef4dd60c5b0b3ac0e03a03d8b61e91","status":"active","name":"discriminative classifiers","description":"Classification methods that learn decision boundaries separating classes.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_20ef4dd60c5b0b3ac0e03a03d8b61e91"},{"public_id":"co_25da511cc3fe94849a0fd49edc02cd1b","status":"active","name":"training algorithms","description":"Procedures used to fit the classifier parameters from data.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_25da511cc3fe94849a0fd49edc02cd1b"},{"public_id":"co_476b9fde7ec7499e8a0368151f60c82e","status":"active","name":"TIMIT speech corpus","description":"An English speech corpus used for evaluating the classifiers.","types":["dataset"],"aliases":["TIMIT"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_476b9fde7ec7499e8a0368151f60c82e"},{"public_id":"co_5e83a1feef46c9123e023fd2182e7733","status":"active","name":"non-linear discriminative classifiers","description":"Discriminative classifiers that allow non-linear decision boundaries.","types":["method"],"aliases":["non-linear classifiers"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_5e83a1feef46c9123e023fd2182e7733"},{"public_id":"co_8039a087448548ed5b6ef6ba33a3bd11","status":"active","name":"linear discriminative classifiers","description":"Discriminative classifiers with linear decision boundaries.","types":["method"],"aliases":["linear classifiers"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_8039a087448548ed5b6ef6ba33a3bd11"},{"public_id":"co_d671f586c1917ceafbfe52dd10da4b89","status":"active","name":"hierarchical tree-root structure","description":"A tree-structured organization applied over phoneme groups in the classifier design.","types":["model structure"],"aliases":["hierarchical tree root structure"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_d671f586c1917ceafbfe52dd10da4b89"},{"public_id":"co_daa38382de406a0fc5efe3f0b849c14d","status":"active","name":"GMM","description":"A Gaussian mixture model used here as a representative generic classifier.","types":["method","baseline"],"aliases":["Gaussian mixture model"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_daa38382de406a0fc5efe3f0b849c14d"},{"public_id":"co_df397595a18a7b530dda1a4713f141e9","status":"active","name":"English language","description":"The language setting of the reported experiments.","types":["language"],"aliases":["English"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_df397595a18a7b530dda1a4713f141e9"},{"public_id":"co_e735fc71e7ad9be3bf42f93dc6dde05d","status":"active","name":"frame-based phoneme classifiers","description":"Classifiers that assign phoneme labels using short-time acoustic frames as input.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_e735fc71e7ad9be3bf42f93dc6dde05d"},{"public_id":"co_f22bdc3f9457c5f66a55c41e04034c8f","status":"active","name":"input phoneme group","description":"The set or grouping of phoneme inputs processed by the classifiers.","types":["input"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_f22bdc3f9457c5f66a55c41e04034c8f"}],"external_ids":{"DOI":"10.15598/AEEE.V9I5.545","ArXiv":null,"PubMed":null,"PubMedCentral":null,"MAG":2073565062,"DBLP":null,"ACL":null},"open_access":{"is_open_access":false,"pdf_url":null,"landing_url":"https://sah.borca.ai/papers/54763794","source":null,"pdf_url_source":null,"license":null,"reason":"pdf_url_not_indexed"},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":false,"full_text_source":null,"count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":631552,"paper_uid":"b1961121-3d0e-49fb-9db4-7f772ef18db0","canonical_identity":{"paper_id":631552,"paper_uid":"b1961121-3d0e-49fb-9db4-7f772ef18db0","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/54763794"}