Papers › SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech

SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech

19 Nov 2021arXiv:2111.10367archive 2025-07-28

Suwon Shon, Ankita Pasad, Felix Wu, Pablo Brusco, Yoav Artzi, Karen Livescu, Kyu J. Han

Progress in speech processing has been facilitated by shared datasets and benchmarks. Historically these have focused on automatic speech recognition (ASR), speaker identification, or other lower-level tasks. Interest has been growing in higher-level spoken language understanding tasks, including using end-to-end models, but there are fewer annotated datasets for such tasks. At the same time, recent work shows the possibility of pre-training generic representations and then fine-tuning for several tasks using relatively little labeled data. We propose to create a suite of benchmark tasks for Spoken Language Understanding Evaluation (SLUE) consisting of limited-size labeled training sets and corresponding evaluation sets. This resource would allow the research community to track progress, evaluate pre-trained representations for higher-level tasks, and study open questions such as the utility of pipeline versus end-to-end approaches. We present the first phase of the SLUE benchmark suite, consisting of named entity recognition, sentiment analysis, and ASR on the corresponding datasets. We focus on naturally produced (not read or synthesized) speech, and freely available datasets. We provide new transcriptions and annotations on subsets of the VoxCeleb and VoxPopuli datasets, evaluation metrics and results for baseline models, and an open-source toolkit to reproduce the baselines and evaluate new models.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 12 code samples harvested from 1 repository linked to this paper: 11 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 12 samples from 1 repository, 11 ran (11 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

asappresearch/slue-toolkit officialmentioned in papermentioned on GitHubpytorchMITSyntology: 12 samples, 11 ran (0 instrument-failed) · 1 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

12 samples harvested: 11 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

11ran
1unverified

Licence: 0 of the 12 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from asappresearch/slue-toolkit. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: 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. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

asappresearch/slue-toolkit official repository 12 samples, 11 ran (0 instrument-failed) · 1 unverified
convert in 76 papers slue_toolkit/eval/eval_utils_nel.py ran fingerprinted MIT (permissive) · 0302aff920740fbc · report fixture: signature sec:float/0/float · returned int · level recorded 2026-09-24
eval_ctc_model slue_toolkit/eval/eval_w2v.py ran MIT (permissive) · 3b9788e17612ec5c · report fixture: signature model:str/0/str, lm:str/0/str, beam_size:int/0/int, lm_weight:float/0/float, word_score:float/0/float, subset:str/0/str, data:str/0/str, upsample:float/0/float, save_results:bool/0/bool, dump_emissions:bool/0/bool, csv_log_file:NoneType/0/?, fp16:bool/0/bool, batch_size:int/0/int, quiet:bool/0/bool, use_bpe:bool/0/bool, user_dir:str/0/str, dry_run:bool/0/bool, max_tokens:int/0/int · returned tuple · level recorded 2026-09-24
evaluate in 514 papers slue_toolkit/eval/eval_utils_nel.py ran fingerprinted MIT (permissive) · c8831cc7145df3ab · report fixture: signature true_pos:int/0/int, false_neg:int/0/int, false_pos:int/0/int · returned tuple · level recorded 2026-09-24
get_ner_scores slue_toolkit/eval/eval_utils_ner.py ran MIT (permissive) · 7df2656f9761c553 · report fixture: signature all_gt:list/1/list, all_predictions:list/1/list · returned dict · level recorded 2026-09-24
get_ner_stats slue_toolkit/eval/eval_utils_ner.py ran MIT (permissive) · 7f746dc6c2a62bb2 · report fixture: signature all_gt:list/1/list, all_predictions:list/1/list · returned dict · level recorded 2026-09-24
load_dct slue_toolkit/generic_utils.py ran MIT (permissive) · 656dc5c1fcd38ff8 · report fixture: signature fname:str/0/str · returned dict · level recorded 2026-09-24
load_json in 344 papers slue_toolkit/generic_utils.py ran MIT (permissive) · 16afefff56a55570 · report fixture: signature fname:str/0/str · returned dict · level recorded 2026-09-24
load_pkl in 46 papers slue_toolkit/generic_utils.py ran MIT (permissive) · e29d3eb7f31a6d05 · report fixture: signature fname:str/0/str, encdng:NoneType/0/? · returned dict · level recorded 2026-09-24
make_distinct slue_toolkit/eval/eval_w2v_ner.py ran MIT (permissive) · 5b6c17ce2ecdeb80 · report fixture: signature label_lst:list/1/tuple · returned list · level recorded 2026-09-24
safe_divide in 19 papers
census: 3 implementations from 5 papers, 2 distinct outputs on one shared input
slue_toolkit/eval/eval_utils_ner.py ran MIT (permissive) · 815aaded7a9eae16 · report fixture: signature numerator:list/1/float, denominator:list/1/float · returned ndarray · level recorded 2026-09-24
undetected_indices slue_toolkit/eval/eval_utils_nel.py ran MIT (permissive) · c6afd1a15980c103 · report fixture: signature gt_tuple_dct:dict/0/?, pred_tuple_dct:dict/0/? · returned list · level recorded 2026-09-24
parse_result in 8 papers slue_toolkit/eval/eval_w2v.py unverified MIT (permissive) · 400008f29f6e7364 · report

Tasks

Automatic Speech RecognitionAutomatic Speech Recognition (ASR)Named Entity RecognitionNamed Entity Recognition (NER)Sentiment AnalysisSpeaker IdentificationSpeech RecognitionSpoken Language Understandingnamed-entity-recognitionspeech-recognition

Datasets

Introduced by this paper, per the archive.

SLUE

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (pipeline approach, uses LM) F1 (%) 69.6 #1 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (pipeline approach, uses LM) Text model DeBERTa-L #1 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (pipeline approach, uses LM) label-F1 (%) 82.2 #1 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (pipeline approach, uses LM) F1 (%) 68.0 #2 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (pipeline approach, uses LM) Text model DeBERTa-L #2 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (pipeline approach, uses LM) label-F1 (%) 79.8 #2 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (e2e approach, uses LM) F1 (%) 64.8 #4 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (e2e approach, uses LM) Text model N/A #4 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (e2e approach, uses LM) label-F1 (%) 73.3 #4 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (e2e approach, uses LM) F1 (%) 63.4 #5 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (e2e approach, uses LM) Text model N/A #5 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (e2e approach, uses LM) label-F1 (%) 71.7 #5 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE HuBERT-B-LS960 (e2e approach, uses LM) F1 (%) 61.9 #6 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE HuBERT-B-LS960 (e2e approach, uses LM) Text model N/A #6 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE HuBERT-B-LS960 (e2e approach, uses LM) label-F1 (%) 70.3 #6 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-VP100K (e2e approach, uses LM) F1 (%) 61.8 #7 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-VP100K (e2e approach, uses LM) Text model N/A #7 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-VP100K (e2e approach, uses LM) label-F1 (%) 69.8 #7 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (pipeline approach) F1 (%) 57.8 #8 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (pipeline approach) Text model DeBERTa-L #8 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (pipeline approach) label-F1 (%) 78.8 #8 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (e2e approach) F1 (%) 50.9 #9 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (e2e approach) Text model - #9 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-L-LL60K (e2e approach) label-F1 (%) 64.7 #9 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (e2e approach) F1 (%) 50.2 #10 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (e2e approach) Text model - #10 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (e2e approach) label-F1 (%) 64.0 #10 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE HuBERT-B-LS960 (e2e approach) F1 (%) 49.8 #11 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE HuBERT-B-LS960 (e2e approach) Text model - #11 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE HuBERT-B-LS960 (e2e approach) label-F1 (%) 62.9 #11 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (pipeline approach) F1 (%) 49.5 #12 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (pipeline approach) Text model DeBERTa-L #12 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-LS960 (pipeline approach) label-F1 (%) 74.2 #12 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-VP100K (e2e approach) F1 (%) 47.9 #13 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-VP100K (e2e approach) Text model - #13 of 13 Archive leaderboard report
Named Entity Recognition (NER) SLUE W2V2-B-VP100K (e2e approach) label-F1 (%) 60.8 #13 of 13 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (pipeline approach, uses LM) F1 (%) 63.3 #1 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (pipeline approach, uses LM) Recall (%) 60.4 #1 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (pipeline approach, uses LM) Text model DeBERTa-L #1 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (pipeline approach) F1 (%) 63.3 #2 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (pipeline approach) Recall (%) 60.2 #2 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (pipeline approach) Text model DeBERTa-L #2 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (pipeline approach, uses LM) F1 (%) 62.9 #3 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (pipeline approach, uses LM) Recall (%) 60.0 #3 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (pipeline approach, uses LM) Text model DeBERTa-L #3 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (pipeline approach) F1 (%) 61.8 #4 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (pipeline approach) Recall (%) 59.0 #4 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (pipeline approach) Text model DeBERTa-L #4 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (e2e approach) F1 (%) 48.5 #5 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (e2e approach) Recall (%) 49.2 #5 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-L-LL60K (e2e approach) Text model N/A #5 of 8 Archive leaderboard report
Sentiment Analysis SLUE HuBERT-B-LS960 (e2e approach) F1 (%) 48.0 #6 of 8 Archive leaderboard report
Sentiment Analysis SLUE HuBERT-B-LS960 (e2e approach) Recall (%) 47.5 #6 of 8 Archive leaderboard report
Sentiment Analysis SLUE HuBERT-B-LS960 (e2e approach) Text model N/A #6 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (e2e approach) F1 (%) 46.6 #7 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (e2e approach) Recall (%) 46.0 #7 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-LS960 (e2e approach) Text model N/A #7 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-VP100K (e2e approach) F1 (%) 38.4 #8 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-VP100K (e2e approach) Recall (%) 38.7 #8 of 8 Archive leaderboard report
Sentiment Analysis SLUE W2V2-B-VP100K (e2e approach) Text model N/A #8 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ TED-LIUM 3 LM) VoxCeleb (Dev) 9.1 #1 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ TED-LIUM 3 LM) VoxCeleb (Test) 10.8 #1 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ TED-LIUM 3 LM) VoxPopuli (Dev) 9.1 #1 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ TED-LIUM 3 LM) VoxPopuli (Test) 9.3 #1 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ TED-LIUM 3 LM) VoxCeleb (Dev) 13.2 #2 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ TED-LIUM 3 LM) VoxCeleb (Test) 15.8 #2 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ TED-LIUM 3 LM) VoxPopuli (Dev) 12.0 #2 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ TED-LIUM 3 LM) VoxPopuli (Test) 12.2 #2 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ in-domain LM) VoxCeleb (Dev) 11.8 #3 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ in-domain LM) VoxCeleb (Test) 13.8 #3 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ in-domain LM) VoxPopuli (Dev) 12.0 #3 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K (+ in-domain LM) VoxPopuli (Test) 12.5 #3 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K VoxCeleb (Dev) 11.0 #4 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K VoxCeleb (Test) 13.5 #4 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K VoxPopuli (Dev) 14.0 #4 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-L-LL60K VoxPopuli (Test) 12.1 #4 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ in-domain LM) VoxCeleb (Dev) 15.2 #5 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ in-domain LM) VoxCeleb (Test) 18.2 #5 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ in-domain LM) VoxPopuli (Dev) 14.6 #5 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 (+ in-domain LM) VoxPopuli (Test) 15.2 #5 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 VoxCeleb (Dev) 17.2 #6 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 VoxCeleb (Test) 20.5 #6 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 VoxPopuli (Dev) 17.2 #6 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-LS960 VoxPopuli (Test) 17.9 #6 of 8 Archive leaderboard report
Speech Recognition SLUE HuBERT-B-LS960 VoxCeleb (Dev) 19.6 #7 of 8 Archive leaderboard report
Speech Recognition SLUE HuBERT-B-LS960 VoxCeleb (Test) 21.2 #7 of 8 Archive leaderboard report
Speech Recognition SLUE HuBERT-B-LS960 VoxPopuli (Dev) 18.6 #7 of 8 Archive leaderboard report
Speech Recognition SLUE HuBERT-B-LS960 VoxPopuli (Test) 19.1 #7 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-VP100K VoxCeleb (Dev) 29.9 #8 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-VP100K VoxCeleb (Test) 33.4 #8 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-VP100K VoxPopuli (Dev) 21.6 #8 of 8 Archive leaderboard report
Speech Recognition SLUE W2V2-B-VP100K VoxPopuli (Test) 22.4 #8 of 8 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections