{"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/rnn-approaches-to-text-normalization-a","title":"RNN Approaches to Text Normalization: A Challenge","arxiv_id":"1611.00068","date":"2016-10-31","proceeding":null,"authors":["Richard Sproat","Navdeep Jaitly"],"abstract":"This paper presents a challenge to the community: given a large corpus of\nwritten text aligned to its normalized spoken form, train an RNN to learn the\ncorrect normalization function. We present a data set of general text where the\nnormalizations were generated using an existing text normalization component of\na text-to-speech system. This data set will be released open-source in the near\nfuture.\n  We also present our own experiments with this data set with a variety of\ndifferent RNN architectures. While some of the architectures do in fact produce\nvery good results when measured in terms of overall accuracy, the errors that\nare produced are problematic, since they would convey completely the wrong\nmessage if such a system were deployed in a speech application. On the other\nhand, we show that a simple FST-based filter can mitigate those errors, and\nachieve a level of accuracy not achievable by the RNN alone.\n  Though our conclusions are largely negative on this point, we are actually\nnot arguing that the text normalization problem is intractable using an pure\nRNN approach, merely that it is not going to be something that can be solved\nmerely by having huge amounts of annotated text data and feeding that to a\ngeneral RNN model. And when we open-source our data, we will be providing a\nnovel data set for sequence-to-sequence modeling in the hopes that the the\ncommunity can find better solutions.\n  The data used in this work have been released and are available at:\nhttps://github.com/rwsproat/text-normalization-data","url_abs":"http://arxiv.org/abs/1611.00068v2","url_pdf":"http://arxiv.org/pdf/1611.00068v2.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":"rnn-approaches-to-text-normalization-a","repo_url":"https://github.com/rwsproat/text-normalization-data","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"text-normalization","task_name":"Text Normalization"},{"task_slug":"text-to-speech","task_name":"Text to Speech"},{"task_slug":"text-to-speech-1","task_name":"text-to-speech"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1611.00068","atlas_url":"https://app.syntology.ai/?focus=1611.00068","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}