{"url":"/sota/speech-recognition-on-tedlium","task":{"name":"Speech Recognition","url":"/task/speech-recognition","note":null},"dataset":{"name":"Tedlium","url":"/dataset/ted-lium-3"},"category":"Audio","categories":["Audio","Speech"],"category_note":null,"description":"**Speech Recognition** is the task of converting spoken language into text. It involves recognizing the words spoken in an audio recording and transcribing them into a written format. The goal is to accurately transcribe the speech in real-time or from recorded audio, taking into account factors such as accents, speaking speed, and background noise.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [SpecAugment](https://arxiv.org/pdf/1904.08779v2.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Word Error Rate (WER)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Word Error Rate (WER)":"lower"}},"counts":{"rows":4,"rows_with_code":1,"rows_with_paper_page":4,"rows_dated":4,"rows_using_additional_data":2},"rows":[{"rank_in_archive_order":1,"model":"United-MedASR (764M)","metrics":{"Word Error Rate (WER)":"0.29"},"uses_additional_data":true,"paper_date":"2024-11-24","paper":"/paper/high-precision-medical-speech-recognition","paper_url":"https://arxiv.org/abs/2412.00055v1","paper_title":"High-precision medical speech recognition through synthetic data and semantic correction: UNITED-MEDASR","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":2,"model":"parakeet-rnnt-1.1b","metrics":{"Word Error Rate (WER)":"3.92"},"uses_additional_data":true,"paper_date":"2023-05-08","paper":"/paper/fast-conformer-with-linearly-scalable","paper_url":"https://arxiv.org/abs/2305.05084v6","paper_title":"Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":3,"model":"Whispering-LLaMa-7b","metrics":{"Word Error Rate (WER)":"4.6"},"uses_additional_data":false,"paper_date":"2023-09-27","paper":"/paper/hyporadise-an-open-baseline-for-generative-1","paper_url":"https://arxiv.org/abs/2309.15701v2","paper_title":"HyPoradise: An Open Baseline for Generative Speech Recognition with Large Language Models","code":"https://github.com/hypotheses-paradise/hypo2trans","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":4,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"SpeechStew (100M)","metrics":{"Word Error Rate (WER)":"5.3"},"uses_additional_data":false,"paper_date":"2021-04-05","paper":"/paper/speechstew-simply-mix-all-available-speech","paper_url":"https://arxiv.org/abs/2104.02133v3","paper_title":"SpeechStew: Simply Mix All Available Speech Recognition Data to Train One Large Neural Network","code":null,"n_code_links":0,"syntology":null}],"since_archive":{"present":false,"note":"No Syntology-extracted rows are published in this build."},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":1,"rows_with_any_sample_ran":1,"distinct_papers_with_graph_line":1,"distinct_papers_with_any_sample_ran":1,"samples_over_distinct_papers":{"n_ran":3,"n_unverified":4,"n_samples":7,"n_pointer_only_licence":0,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":3,"n_unverified":4,"n_samples":7,"n_pointer_only_licence":0,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}