{"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/language-identification-using-deep","title":"Language Identification Using Deep Convolutional Recurrent Neural Networks","arxiv_id":"1708.04811","date":"2017-08-16","proceeding":null,"authors":["Christian Bartz","Tom Herold","Haojin Yang","Christoph Meinel"],"abstract":"Language Identification (LID) systems are used to classify the spoken\nlanguage from a given audio sample and are typically the first step for many\nspoken language processing tasks, such as Automatic Speech Recognition (ASR)\nsystems. Without automatic language detection, speech utterances cannot be\nparsed correctly and grammar rules cannot be applied, causing subsequent speech\nrecognition steps to fail. We propose a LID system that solves the problem in\nthe image domain, rather than the audio domain. We use a hybrid Convolutional\nRecurrent Neural Network (CRNN) that operates on spectrogram images of the\nprovided audio snippets. In extensive experiments we show, that our model is\napplicable to a range of noisy scenarios and can easily be extended to\npreviously unknown languages, while maintaining its classification accuracy. We\nrelease our code and a large scale training set for LID systems to the\ncommunity.","url_abs":"http://arxiv.org/abs/1708.04811v1","url_pdf":"http://arxiv.org/pdf/1708.04811v1.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":"language-identification-using-deep","repo_url":"https://github.com/HPI-DeepLearning/crnn-lid","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"automatic-speech-recognition-2","task_name":"Automatic Speech Recognition"},{"task_slug":"automatic-speech-recognition","task_name":"Automatic Speech Recognition (ASR)"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"language-identification","task_name":"Language Identification"},{"task_slug":"speech-recognition","task_name":"Speech Recognition"},{"task_slug":"spoken-language-identification","task_name":"Spoken language identification"},{"task_slug":"speech-recognition-1","task_name":"speech-recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/spoken-language-identification-on-youtube-3","task":"Spoken language identification","dataset":"YouTube News dataset (Background Music)","model":"Inception-v3 CRNN","rank_in_archive_order":1,"of":2,"metrics":{"Accuracy ":"0.89","F1 Score":"0.89"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube-3","task":"Spoken language identification","dataset":"YouTube News dataset (Background Music)","model":"CRNN","rank_in_archive_order":2,"of":2,"metrics":{"Accuracy ":"0.70","F1 Score":"0.70"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube-2","task":"Spoken language identification","dataset":"YouTube News dataset (Crackling Noise)","model":"Inception-v3 CRNN","rank_in_archive_order":1,"of":2,"metrics":{"Accuracy ":"0.93","F1 Score":"0.93"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube-2","task":"Spoken language identification","dataset":"YouTube News dataset (Crackling Noise)","model":"CRNN","rank_in_archive_order":2,"of":2,"metrics":{"Accuracy ":"0.82","F1 Score":"0.83"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube","task":"Spoken language identification","dataset":"YouTube News dataset (No Noise)","model":"Inception-v3 CRNN","rank_in_archive_order":3,"of":5,"metrics":{"Accuracy ":"0.96","F1 Score":"0.96"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube","task":"Spoken language identification","dataset":"YouTube News dataset (No Noise)","model":"CRNN","rank_in_archive_order":5,"of":5,"metrics":{"Accuracy ":"0.91","F1 Score":"0.91"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube-1","task":"Spoken language identification","dataset":"YouTube News dataset (White Noise)","model":"Inception-v3 CRNN","rank_in_archive_order":2,"of":5,"metrics":{"Accuracy ":"0.91","F1 Score":"0.91"},"uses_additional_data":false},{"leaderboard":"/sota/spoken-language-identification-on-youtube-1","task":"Spoken language identification","dataset":"YouTube News dataset (White Noise)","model":"CRNN","rank_in_archive_order":5,"of":5,"metrics":{"Accuracy ":"0.63","F1 Score":"0.63"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}