{"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/improved-training-of-end-to-end-attention","title":"Improved training of end-to-end attention models for speech recognition","arxiv_id":"1805.03294","date":"2018-05-08","proceeding":null,"authors":["Albert Zeyer","Kazuki Irie","Ralf Schlüter","Hermann Ney"],"abstract":"Sequence-to-sequence attention-based models on subword units allow simple\nopen-vocabulary end-to-end speech recognition. In this work, we show that such\nmodels can achieve competitive results on the Switchboard 300h and LibriSpeech\n1000h tasks. In particular, we report the state-of-the-art word error rates\n(WER) of 3.54% on the dev-clean and 3.82% on the test-clean evaluation subsets\nof LibriSpeech. We introduce a new pretraining scheme by starting with a high\ntime reduction factor and lowering it during training, which is crucial both\nfor convergence and final performance. In some experiments, we also use an\nauxiliary CTC loss function to help the convergence. In addition, we train long\nshort-term memory (LSTM) language models on subword units. By shallow fusion,\nwe report up to 27% relative improvements in WER over the attention baseline\nwithout a language model.","url_abs":"http://arxiv.org/abs/1805.03294v1","url_pdf":"http://arxiv.org/pdf/1805.03294v1.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":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/rwth-i6/returnn","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/rwth-i6/returnn-experiments","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/bobchennan/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/creatorscan/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/danoneata/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/dhanya-e/Google_Indic","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/jumon/espnet-1","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/jzmo/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/marynader6/espnetV.8","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/marynader6/transformerTTS","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/pengchengguo/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/pzelasko/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/victor45664/espnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improved-training-of-end-to-end-attention","repo_url":"https://github.com/vinitunni/CoupledLoss-LAS-ESPNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"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"}],"methods":[{"method_slug":"ctc-loss","method_name":"CTC Loss"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/speech-recognition-on-librispeech-test-clean","task":"Speech Recognition","dataset":"LibriSpeech test-clean","model":"Seq-to-seq attention","rank_in_archive_order":51,"of":64,"metrics":{"Word Error Rate (WER)":"3.82"},"uses_additional_data":true}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1805.03294","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1805.03294"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/bobchennan/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/pzelasko/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jumon/espnet-1","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/pengchengguo/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/rwth-i6/returnn","reach":{"status":"ok","spdx":"NOASSERTION"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/marynader6/espnetV.8","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/creatorscan/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/danoneata/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/dhanya-e/Google_Indic","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/vinitunni/CoupledLoss-LAS-ESPNet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/marynader6/transformerTTS","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/rwth-i6/returnn-experiments","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jzmo/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/victor45664/espnet","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"unverified":12},"by_repo_kind":{"listed":{"samples":12,"ran":0,"repositories":2}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"bbe192baf1d74dd1","entry":"add_results_to_json","repo":"creatorscan/espnet","repo_kind":"listed","path":"espnet/mt/mt_utils.py","file_url":"https://github.com/creatorscan/espnet/blob/HEAD/espnet/mt/mt_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"bbe192baf1d74dd1"}},{"code_sha256_prefix":"ff899e7de6121891","entry":"count_tokens","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/lm/lm_utils.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/lm/lm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ff899e7de6121891"}},{"code_sha256_prefix":"e2309fb95e3f748c","entry":"count_tokens","repo":"creatorscan/espnet","repo_kind":"listed","path":"espnet/lm/lm_utils.py","file_url":"https://github.com/creatorscan/espnet/blob/HEAD/espnet/lm/lm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e2309fb95e3f748c"}},{"code_sha256_prefix":"db10d7f5f75d9faa","entry":"end_detect","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/nets/e2e_asr_common.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/nets/e2e_asr_common.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"db10d7f5f75d9faa"}},{"code_sha256_prefix":"34e1e172156ca868","entry":"get_vgg2l_odim","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/nets/e2e_asr_common.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/nets/e2e_asr_common.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"34e1e172156ca868"}},{"code_sha256_prefix":"0d9f822390e7d551","entry":"label_smoothing_dist","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/nets/e2e_asr_common.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/nets/e2e_asr_common.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0d9f822390e7d551"}},{"code_sha256_prefix":"826840e1c3869a70","entry":"load_dataset","repo":"creatorscan/espnet","repo_kind":"listed","path":"espnet/lm/lm_utils.py","file_url":"https://github.com/creatorscan/espnet/blob/HEAD/espnet/lm/lm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"826840e1c3869a70"}},{"code_sha256_prefix":"6de087d747415428","entry":"make_batchset","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/asr/asr_mix_utils.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/asr/asr_mix_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"6de087d747415428"}},{"code_sha256_prefix":"2956cc98fe367b81","entry":"make_lexical_tree","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/lm/lm_utils.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/lm/lm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"2956cc98fe367b81"}},{"code_sha256_prefix":"48cef9e2f067bc77","entry":"parse_hypothesis","repo":"creatorscan/espnet","repo_kind":"listed","path":"espnet/mt/mt_utils.py","file_url":"https://github.com/creatorscan/espnet/blob/HEAD/espnet/mt/mt_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"48cef9e2f067bc77"}},{"code_sha256_prefix":"b9cca15d6b089537","entry":"read_tokens","repo":"bobchennan/espnet","repo_kind":"listed","path":"espnet/lm/lm_utils.py","file_url":"https://github.com/bobchennan/espnet/blob/HEAD/espnet/lm/lm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b9cca15d6b089537"}},{"code_sha256_prefix":"dc5fc4e9e58970cb","entry":"read_tokens","repo":"creatorscan/espnet","repo_kind":"listed","path":"espnet/lm/lm_utils.py","file_url":"https://github.com/creatorscan/espnet/blob/HEAD/espnet/lm/lm_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"dc5fc4e9e58970cb"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}