{"id":238104,"date":"2016-03-01T00:00:00","date_gmt":"2016-03-01T08:00:00","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/msr-research-item\/exploring-multidimensional-lstms-for-large-vocabulary-asr\/"},"modified":"2018-10-16T19:58:36","modified_gmt":"2018-10-17T02:58:36","slug":"exploring-multidimensional-lstms-for-large-vocabulary-asr","status":"publish","type":"msr-research-item","link":"https:\/\/www.microsoft.com\/en-us\/research\/publication\/exploring-multidimensional-lstms-for-large-vocabulary-asr\/","title":{"rendered":"Exploring Multidimensional LSTMs for Large Vocabulary ASR"},"content":{"rendered":"<div class=\"asset-content\">\n<p>Long short-term memory (LSTM) recurrent neural networks (RNNs) have recently shown significant performance improvements over deep feed-forward neural networks. A key aspect of these models is the use of time recurrence, combined with a gating architecture that allows them to track the long-term dynamics of speech. Inspired by human spectrogram reading, we recently proposed the frequency LSTM (F-LSTM) that performs 1-D recurrence over the frequency axis and then performs 1-D recurrence over the time axis. In this study, we further improve the acoustic model by proposing a 2-D, time-frequency (TF) LSTM. The TF-LSTM jointly scans the input over the time and frequency axes to model spectro-temporal warping, and then uses the output activations as the input to a time LSTM (T-LSTM). The joint time-frequency modeling better normalizes the features for the upper layer T-LSTMs. Evaluated on a 375-hour short message dictation task, the proposed TF-LSTM obtained a 3.4% relative WER reduction over the best T-LSTM. The invariance property achieved by joint time-frequency analysis is demonstrated on a mismatched test set, where the TF-LSTM achieves a 14.2% relative WER reduction over the best T-LSTM.<\/p>\n<\/div>\n<p><!-- .asset-content --><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Long short-term memory (LSTM) recurrent neural networks (RNNs) have recently shown significant performance improvements over deep feed-forward neural networks. A key aspect of these models is the use of time recurrence, combined with a gating architecture that allows them to track the long-term dynamics of speech. Inspired by human spectrogram reading, we recently proposed the [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13547],"msr-publication-type":[193716],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-238104","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-systems-and-networking","msr-locale-en_us"],"msr_publishername":"","msr_edition":"ICASSP","msr_affiliation":"","msr_published_date":"2016-03-01","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"ICASSP","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"238415","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"file","title":"TFLSTM.pdf","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2016\/06\/TFLSTM-1.pdf","id":238415,"label_id":0}],"msr_related_uploader":"","msr_attachments":[{"id":238415,"url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2016\/06\/TFLSTM-1.pdf"}],"msr-author-ordering":[{"type":"text","value":"Jinyu Li","user_id":0,"rest_url":false},{"type":"user_nicename","value":"abmoh","user_id":30798,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=abmoh"},{"type":"user_nicename","value":"gzweig","user_id":31938,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=gzweig"},{"type":"user_nicename","value":"ygong","user_id":34994,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=ygong"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"download":[],"msr_publication_type":"inproceedings","related_content":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/238104","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":1,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/238104\/revisions"}],"predecessor-version":[{"id":515957,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/238104\/revisions\/515957"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=238104"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=238104"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=238104"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=238104"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=238104"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=238104"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=238104"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=238104"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=238104"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=238104"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=238104"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=238104"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=238104"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=238104"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=238104"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=238104"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}