{"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/segmental-recurrent-neural-networks-for-end","title":"Segmental Recurrent Neural Networks for End-to-end Speech Recognition","arxiv_id":"1603.00223","date":"2016-03-01","proceeding":null,"authors":["Liang Lu","Lingpeng Kong","Chris Dyer","Noah A. Smith","Steve Renals"],"abstract":"We study the segmental recurrent neural network for end-to-end acoustic\nmodelling. This model connects the segmental conditional random field (CRF)\nwith a recurrent neural network (RNN) used for feature extraction. Compared to\nmost previous CRF-based acoustic models, it does not rely on an external system\nto provide features or segmentation boundaries. Instead, this model\nmarginalises out all the possible segmentations, and features are extracted\nfrom the RNN trained together with the segmental CRF. In essence, this model is\nself-contained and can be trained end-to-end. In this paper, we discuss\npractical training and decoding issues as well as the method to speed up the\ntraining in the context of speech recognition. We performed experiments on the\nTIMIT dataset. We achieved 17.3 phone error rate (PER) from the first-pass\ndecoding --- the best reported result using CRFs, despite the fact that we only\nused a zeroth-order CRF and without using any language model.","url_abs":"http://arxiv.org/abs/1603.00223v2","url_pdf":"http://arxiv.org/pdf/1603.00223v2.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":[],"tasks":[{"task_slug":"acoustic-modelling","task_name":"Acoustic Modelling"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"speech-recognition","task_name":"Speech Recognition"},{"task_slug":"speech-recognition-1","task_name":"speech-recognition"}],"methods":[{"method_slug":"crf","method_name":"CRF"},{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/speech-recognition-on-timit","task":"Speech Recognition","dataset":"TIMIT","model":"RNN-CRF on 24(x3) MFSC","rank_in_archive_order":16,"of":22,"metrics":{"Percentage error":"17.3"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1603.00223","atlas_url":"https://app.syntology.ai/?focus=1603.00223","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}