{"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/the-microsoft-2016-conversational-speech","title":"The Microsoft 2016 Conversational Speech Recognition System","arxiv_id":"1609.03528","date":"2016-09-12","proceeding":null,"authors":["W. Xiong","J. Droppo","X. Huang","F. Seide","M. Seltzer","A. Stolcke","D. Yu","G. Zweig"],"abstract":"We describe Microsoft's conversational speech recognition system, in which we\ncombine recent developments in neural-network-based acoustic and language\nmodeling to advance the state of the art on the Switchboard recognition task.\nInspired by machine learning ensemble techniques, the system uses a range of\nconvolutional and recurrent neural networks. I-vector modeling and lattice-free\nMMI training provide significant gains for all acoustic model architectures.\nLanguage model rescoring with multiple forward and backward running RNNLMs, and\nword posterior-based system combination provide a 20% boost. The best single\nsystem uses a ResNet architecture acoustic model with RNNLM rescoring, and\nachieves a word error rate of 6.9% on the NIST 2000 Switchboard task. The\ncombined system has an error rate of 6.2%, representing an improvement over\npreviously reported results on this benchmark task.","url_abs":"http://arxiv.org/abs/1609.03528v2","url_pdf":"http://arxiv.org/pdf/1609.03528v2.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":"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":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"bottleneck-residual-block","method_name":"Bottleneck Residual Block"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"kaiming-initialization","method_name":"Kaiming Initialization"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"residual-block","method_name":"Residual Block"},{"method_slug":"residual-connection","method_name":"Residual Connection"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/speech-recognition-on-switchboard-hub500","task":"Speech Recognition","dataset":"Switchboard + Hub500","model":"Microsoft 2016","rank_in_archive_order":5,"of":30,"metrics":{"Percentage error":"6.2"},"uses_additional_data":false},{"leaderboard":"/sota/speech-recognition-on-switchboard-hub500","task":"Speech Recognition","dataset":"Switchboard + Hub500","model":"VGG/Resnet/LACE/BiLSTM acoustic model trained on SWB+Fisher+CH, N-gram + RNNLM language model trained on Switchboard+Fisher+Gigaword+Broadcast","rank_in_archive_order":6,"of":30,"metrics":{"Percentage error":"6.3"},"uses_additional_data":false},{"leaderboard":"/sota/speech-recognition-on-switchboard-hub500","task":"Speech Recognition","dataset":"Switchboard + Hub500","model":"RNNLM","rank_in_archive_order":10,"of":30,"metrics":{"Percentage error":"6.9"},"uses_additional_data":false},{"leaderboard":"/sota/speech-recognition-on-swb_hub_500-wer","task":"Speech Recognition","dataset":"swb_hub_500 WER fullSWBCH","model":"VGG/Resnet/LACE/BiLSTM acoustic model trained on SWB+Fisher+CH, N-gram + RNNLM language model trained on Switchboard+Fisher+Gigaword+Broadcast","rank_in_archive_order":4,"of":12,"metrics":{"Percentage error":"11.9"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1609.03528","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}