Pith. sign in

REVIEW 4 major objections 4 minor 2 cited by

How "Real" is Your Real-Time Simultaneous Speech-to-Text Translation System?

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that most simultaneous speech-to-text translation research evaluates on human-pre-segmented audio and rarely acknowledges this, so reported results may not transfer to real-world continuous speech.

desk verdict Useful survey, plausible central claim, but the headline percentages need a reproducible coding before they can be trusted. read the letter →

arxiv 2412.18495 v1 pith:EZEZ5JW2 submitted 2024-12-24 cs.CL cs.AIcs.SDeess.AS

Sara Papi , Peter Polak , Ondřej Bojar , Dominik Macháček This is my paper · ORCID
classification cs.CLcs.AIcs.SDeess.AS
keywords simultaneousspeechtranslationspeech-to-textliteraturereviewaudiosegmentationstreamingterminologystandardizationlatencyevaluationunbounded
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that the field of simultaneous speech-to-text translation (SimulST) has drifted from its stated goal of translating a continuous, unsegmented audio stream. The authors surveyed 110 papers and found that up to 81.8% of them operate only on pre-segmented audio, and 97.7% of those rely on gold (human-made) segment boundaries that do not exist in real-time use. They also found that over 65% of papers use "streaming," "online," or "real-time" interchangeably with "simultaneous" without defining these terms. To address these problems, the paper proposes a standardized taxonomy and a six-step formalization of the SimulST process, along with concrete recommendations for evaluation and reporting. If the survey's statistics are correct, much of the reported progress in SimulST may simply not carry over to the real-world conditions the task is meant for.

What carries the argument

The central mechanism is a taxonomy built on three dichotomies: input (bounded vs unbounded speech), architecture (direct vs cascade), and output strategy (incremental vs re-translation), which the authors apply by hand to classify all 110 surveyed papers. Supporting this is a six-step decomposition of the SimulST process, from audio acquisition through buffer updating, hypothesis generation, buffer trimming, and output presentation. The taxonomy does the work of converting a loosely defined body of literature into countable categories, which is what makes the headline percentages (81.8%, 97.7%, 91.8%) possible at all.

What would settle it

Have two independent annotators re-classify the same 110 papers using the taxonomy from Appendix A and measure inter-annotator agreement; if agreement is low (e.g., Cohen's kappa below 0.6) or the re-derived percentages differ materially from the reported 81.8% and 91.8%, the paper's key statistics would not survive.

Watch

Extended reading notes

Core claim

The paper's central claim is that the SimulST community has been evaluating its models under unrealistic input conditions while rarely acknowledging it. Its key quantitative finding is that, among the 110 reviewed papers, up to 81.8% rely on pre-segmented audio, with 97.7% of those using gold (human) segmentation, and 91.8% of all papers do not explicitly state that they assume gold pre-segmented speech. Only 20 papers address unbounded speech at all, and only two explore replacing gold with automatic segmentation in the bounded scenario. The paper also documents terminological chaos, with over 65% of papers mixing at least one of "streaming," "online," or "real-time" with "simultaneous" without clear definitions. As a remedy, it formalizes SimulST as a six-step process and introduces a taxonomy of system components (input type, architecture, output strategy) to make the field's assumptions visible.

Load-bearing premise

The survey's percentages are produced by the authors' manual classification of the 110 papers, so the central claim stands or falls with whether those classifications are accurate and reproducible.

Editorial extensions

If this is right

  • SimulST results obtained on gold-segmented benchmarks should not be treated as predictive of performance on unbounded, continuous audio streams.
  • The dominant evaluation toolkit, SimulEval, needs extensions or a successor that can score systems on streams without relying on pre-segmented inputs.
  • Researchers working with bounded speech inputs should adopt automatic pre-segmentation instead of gold segmentation to more closely approximate real conditions.
  • Adopting a unified terminology and explicitly stating the type of speech input in every paper would make results across the field comparable.
  • Human-centered evaluation of output visualization and of the quality-latency trade-off is needed before automatic metric improvements can be trusted to improve user experience.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct test the paper leaves implicit would be to run the same SimulST model on continuous audio and on gold-segmented audio; a large quality drop on the continuous stream would confirm that the field's bounded-input focus is the main gap.
  • The taxonomy could double as a reporting checklist for future papers, and venues could require authors to state input type explicitly; this is a policy implication the paper hints at but does not spell out.
  • Because the terminology problem is inherited from neighboring fields (ASR's "streaming" and MT's "online"), a durable fix likely requires coordinated glossary efforts across communities rather than a single survey's definitions.
  • The 97.7% gold-segmentation statistic also implies that segmentation algorithms have been evaluated only indirectly, so a dedicated segmentation-aware benchmark could turn the survey's critique into a shared task.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper surveys 110 papers on simultaneous speech-to-text translation (SimulST), proposes a six-step process model and a standardized terminology/taxonomy for the task, and then uses the survey to argue that the field has largely evaluated systems on pre-segmented gold audio rather than on realistic unbounded speech streams. The headline empirical claims in §4 are that 81.8% of papers rely on pre-segmented audio, 97.7% of those use gold segmentation, and 91.8% do not explicitly acknowledge the gold-segmentation assumption. The paper also reports widespread terminological confusion among "simultaneous", "streaming", "online", and "real-time", and it closes with recommendations on automatic segmentation, latency reporting, evaluation frameworks for unbounded speech, context handling, output visualization, and user-centric evaluation.

Significance. If the survey statistics are correct, the paper would provide an important field-level correction: SimulST research would largely be optimizing under an unrealistic input assumption, and reported results would not transfer to continuous speech streams. The conceptual contributions—the six-step process decomposition, the taxonomy in Figure 2, and the terminology in Table 1—are genuinely useful and are likely to be adopted regardless of the exact percentages. The paper is also transparent in providing the full categorized list in Appendix A, which is a strength for reproducibility. However, the empirical claims are load-bearing for the paper's central message, and, as detailed below, the appendix as printed does not yet allow an independent reader to verify the headline percentages.

major comments (4)
  1. [§4 and Appendix A] The headline statistics in §4 are not reproducible from Appendix A as printed. The appendix reports 90 bounded + 20 unbounded + 1 undefined = 111 papers, although the text states 110; the entry "Polák et al. (2023)" appears twice in the same category list while the citation does not distinguish between the two distinct 2023 Polák et al. papers; and the category in A.4 is "Papers Mentioning Automatic Segmentation", not "papers that explicitly acknowledge gold pre-segmentation", so the 91.8% figure in §4 cannot be checked against the appendix. The authors should provide a machine-readable one-row-per-paper table, a precise definition of each variable used in the percentages, and a coding rubric.
  2. [A.1.1] The classification involves judgment calls that are not subjected to any sensitivity analysis. For example, Ma et al. (2021) is listed under gold pre-segmentation with the footnote "Unbounded speech theoretically possible but not tested", and many papers do not state their input conditions explicitly. Since the central claim is that "up to 81.8%" of papers rely on pre-segmented audio and "97.7%" of those use gold segmentation, the authors should report inter-annotator agreement on a sample and show how the percentages shift when boundary cases such as Ma et al. (2021) are reclassified.
  3. [Appendix A, selection criteria] The sample is restricted to open-access English papers retrieved from Semantic Scholar with no flow diagram. The queries in Table 2 each returned between 69 and 265 papers, and the final set of 110 papers is obtained after unspecified deduplication and filtering. This selection could bias the field-level percentages. The authors should report the number of unique papers per query, the exclusion counts, and the exact search date, or temper the conclusions from "the field" to "the open-access English literature surveyed".
  4. [§4, terminological chaos] The claim that "over 65% of the papers mix and match these terms" is not supported by the appendix. No per-paper term-usage data are provided, and Figure 3 is referenced without showing the underlying computation or a table of which terms each paper uses. The authors should include the term-usage data or explicitly define the numerator and denominator behind the 65% figure.
minor comments (4)
  1. [§5, first paragraph] The first paragraph of §5 contains two empty parentheticals, "( )", where references to recent advances and future directions are missing; these need to be filled in.
  2. [A.1.2] The label "Simultaneous (Automatic) Segmentation" is confusing because it can be read as "simultaneous segmentation" rather than "automatic segmentation performed in real time"; consider aligning the label with the taxonomy terms from Table 1 and Figure 2.
  3. [References] Two distinct 2023 papers by Polák et al. are cited in the appendix as identical "Polák et al. (2023)" entries; the authors should disambiguate them, for example as Polák et al. (2023a) and Polák et al. (2023b), so that the counts in Appendix A can be verified.
  4. [Figure 2 and Appendix A] The paper would benefit from a table mapping each Appendix A category name to the corresponding taxonomy node in Figure 2; the current labels such as "Bounded Speech" and "Unbounded Speech" are close to but not identical with the terminology proposed in Table 1.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline percentages are empirical codings of the surveyed literature, not consequences of the paper's own taxonomy.

full rationale

This paper is a literature survey and position paper. Its central claims—that 81.8% of surveyed papers use pre-segmented audio, that 97.7% of those use gold segmentation, and that 91.8% do not explicitly acknowledge the assumption—are empirical tallies over the manual categorizations in Appendix A. These percentages are not entailed by the taxonomy proposed in Section 3: the taxonomy defines categories such as 'bounded' and 'gold pre-segmentation', but assigning each paper to those categories requires reading the papers and is not a formal derivation. The taxonomy is the authors' own construction and is used as an organizing scheme, but the headline statistics could in principle come out differently under the same taxonomy if the coding changed; they are observations, not tautologies. The paper's self-citations (Polák et al. 2023; Papi et al. 2024b, among others) are used as examples of unbounded-speech systems and as prior work on metrics, but they are not the load-bearing evidence for the field-wide statistics; the statistics come from the independent, though manual, coding of the full corpus. There is no equation in the paper that reduces a predicted quantity to a fitted input, no uniqueness theorem imported from the authors' prior work, and no ansatz smuggled in via citation. The Appendix tally is internally inconsistent (90 bounded + 20 unbounded + 1 undefined = 111, while the paper says 110) and the coding rubric is not fully specified; these are reproducibility and correctness concerns, not circularity, and should be addressed with an independent re-coding, but they do not make the derivation circular. Finding: no significant circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The paper introduces no fitted parameters or invented entities. It rests on domain assumptions about sample representativeness and the reliability of the authors' manual paper categorization, plus the framing that unbounded speech is the real-world target.

assumptions (3)
  • domain assumption Semantic Scholar query set and inclusion criteria (English, open-access, model papers with speech-to-text results) yield a representative sample of SimulST research.
    The survey's quantitative claims (e.g., 81.8% use pre-segmented audio) are derived from the 110 retrieved papers. If the open-access or English filters bias the sample, the percentages may not generalize. This is disclosed in Appendix A but not validated.
  • domain assumption Manual classification of each paper into the taxonomy (bounded/unbounded, gold/automatic segmentation, direct/cascade, incremental/re-translation) is accurate and reproducible.
    No coding rubric, inter-annotator agreement, or sensitivity analysis is provided. The lists in Appendix A are the sole evidence for the statistics; a few misclassifications could change the headline percentages.
  • domain assumption The taxonomy's dichotomy between bounded and unbounded speech captures the practically relevant distinction for real-world deployment.
    The paper assumes gold pre-segmented input is unrealistic and unbounded streaming is the target scenario. This framing motivates the recommendations and is a value judgment about what the field should prioritize.

how reviews work

0 comments
Cite this review

Pith. "Pith review of How "Real" is Your Real-Time Simultaneous Speech-to-Text Translation System?." pith.science (2026). https://pith.science/paper/EZEZ5JW2

@misc{pith2026241218495,
  author       = {Pith},
  title        = {Pith review of: How "Real" is Your Real-Time Simultaneous Speech-to-Text Translation System?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EZEZ5JW2}},
  note         = {Machine review of arXiv:2412.18495}
}
read the original abstract

Simultaneous speech-to-text translation (SimulST) translates source-language speech into target-language text concurrently with the speaker's speech, ensuring low latency for better user comprehension. Despite its intended application to unbounded speech, most research has focused on human pre-segmented speech, simplifying the task and overlooking significant challenges. This narrow focus, coupled with widespread terminological inconsistencies, is limiting the applicability of research outcomes to real-world applications, ultimately hindering progress in the field. Our extensive literature review of 110 papers not only reveals these critical issues in current research but also serves as the foundation for our key contributions. We 1) define the steps and core components of a SimulST system, proposing a standardized terminology and taxonomy; 2) conduct a thorough analysis of community trends, and 3) offer concrete recommendations and future directions to bridge the gaps in existing literature, from evaluation frameworks to system architectures, for advancing the field towards more realistic and effective SimulST solutions.

Figures

Figures reproduced from arXiv: 2412.18495 by the authors.

Figure 1
Figure 1. Representation of the steps (1 to 6) of the SimulST process. 5. (Optional) Speech and Text Buffers Trim￾ming: The content of the Speech and Text Buffers (BS and BT) is trimmed based on the audio-textual information to be retained from the past. This step makes the size of the buffers manageable by ST models, which cannot deal with an infinitely growing con￾text. The content is determined by a trim function, which ke… view at source ↗
Figure 2
Figure 2. Taxonomy of the SimulST solutions. employ separate ASR and MT systems, where the best hypothesis of the former serves as in￾put to the latter. Bahar et al. (2019) surveyed various direct architectures, many of which lever￾age multi-task training (Luong et al., 2016) – e.g., incorporating Connectionist Temporal Classifica￾tion (CTC) loss computed on transcripts (Graves et al., 2006) alongside standard cross-entropy l… view at source ↗
Figure 3
Figure 3. Waffle plot of the term “simultane￾ous” and commonly used synonyms (“streaming”, “real-time”, and “online”) among the 110 catego￾rized papers. tion task, further adding to the confusion. This in￾consistent terminology creates significant ambigu￾ity, making it challenging to understand the tasks being addressed, especially when terms are used without explicit definitions. The lack of unifor￾mity calls for a clear, co… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Number of papers in our survey employing direct or cascade simultaneous ST architectures [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Prepending or Cross-Attention for Speech-to-Text? An Empirical Comparison

    cs.CL 2025-01 conditional novelty 5.0 of 10

    Across controlled ASR and speech translation experiments, dense feature prepending does not outperform cross-attention in quality and is slightly slower and more memory hungry.

  2. Streaming Speaker Change Detection and Gender Classification for Transducer-Based Multi-Talker Speech Translation

    cs.SD 2025-02 conditional novelty 4.0 of 10

    Cosine similarity between t-vector speaker embeddings in a streaming transducer speech translation model detects speaker changes (F1 up to 0.68) and classifies gender (0.989 accuracy).

Reference graph

Works this paper leans on

31 extracted references · 23 canonical work pages · cited by 2 Pith papers

  1. [5]

    Association for Computational Linguistics

    In Proceedings of the 20th Interna- tional Conference on Spoken Language Trans- lation (IWSLT 2023), pages 330–340, Toronto, Canada (in-person and online). Association for Computational Linguistics. Ryo Fukuda, Katsuhito Sudoh, and Satoshi Naka- mura. 2022b. Speech Segmentation Optimiza- tion using Segmented Bilingual Speech Corpus for End-to-end Speech T...

    work page 2023

  2. [8]

    In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2599–2611, Online

    Direct segmentation models for stream- ing speech translation. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2599–2611, Online. Association for Computa- tional Linguistics. Javier Iranzo-Sánchez, Javier Jorge Cano, Ale- jandro Pérez-González-de Martos, Adrián Giménez Pastor, Gonçal Garcés Díaz-Mun...

    work page 2020

  3. [10]

    Association for Computational Linguistics

    When and why is document-level context useful in neural machine translation? In Pro- ceedings of the Fourth Workshop on Discourse in Machine Translation (DiscoMT 2019) , page 24–34, Hong Kong, China. Association for Computational Linguistics. Yuka Ko, Ryo Fukuda, Yuta Nishikawa, Yasumasa Kano, Katsuhito Sudoh, and Satoshi Nakamura

    work page 2019

  4. [11]

    In Proceedings of the 20th Interna- tional Conference on Spoken Language Trans- lation (IWSLT 2023), pages 363–375, Toronto, Canada (in-person and online)

    Tagged end-to-end simultaneous speech translation training using simultaneous interpre- tation data. In Proceedings of the 20th Interna- tional Conference on Spoken Language Trans- lation (IWSLT 2023), pages 363–375, Toronto, Canada (in-person and online). Association for Computational Linguistics. Yuka Ko, Ryo Fukuda, Yuta Nishikawa, Yasumasa Kano, Tomoy...

    work page 2023

  5. [12]

    In Proceedings of the 21st International Conference on Spoken Lan- guage Translation (IWSLT 2024) , pages 170– 182, Bangkok, Thailand (in-person and online)

    NAIST simultaneous speech translation system for IWSLT 2024. In Proceedings of the 21st International Conference on Spoken Lan- guage Translation (IWSLT 2024) , pages 170– 182, Bangkok, Thailand (in-person and online). Association for Computational Linguistics. Tom Kocmi, Vilém Zouhar, Christian Federmann, and Matt Post. 2024. Navigating the met- rics maz...

    work page 2024

  6. [14]

    In Proceedings of the 62nd Annual Meeting of the Association for Compu- tational Linguistics (Volume 1: Long Papers) , pages 1557–1575, Bangkok, Thailand

    A non-autoregressive generation frame- work for end-to-end simultaneous speech-to- any translation. In Proceedings of the 62nd Annual Meeting of the Association for Compu- tational Linguistics (Volume 1: Long Papers) , pages 1557–1575, Bangkok, Thailand. Associ- ation for Computational Linguistics. Dominik Macháˇcek, Ondˇrej Bojar, and Raj Dabre

    work page

  7. [15]

    In Proceedings of the 20th International Confer- ence on Spoken Language Translation (IWSLT 2023), pages 169–179, Toronto, Canada (in- person and online)

    MT metrics correlate with human rat- ings of simultaneous speech translation. In Proceedings of the 20th International Confer- ence on Spoken Language Translation (IWSLT 2023), pages 169–179, Toronto, Canada (in- person and online). Association for Computa- tional Linguistics. Dominik Machá ˇcek, Jonáš Kratochvíl, Sangeet Sagar, Matúš Žilinec, Ond ˇrej Bo...

    work page 2023

  8. [16]

    In ICASSP 2020 - 2020 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP) , pages 6074–6078

    Streaming automatic speech recognition with the transformer model. In ICASSP 2020 - 2020 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP) , pages 6074–6078. Markus Müller, Thai Son Nguyen, Jan Niehues, Eunah Cho, Bastian Krüger, Thanh-Le Ha, Kevin Kilgour, Matthias Sperber, Mohammed Mediani, Sebastian Stüker, and Alex Waibel

    work page 2020

Show all 31 references
  1. [18]

    In Proceedings of the 16th International Con- ference on Spoken Language Translation, Hong Kong

    The IWSLT 2019 evaluation campaign. In Proceedings of the 16th International Con- ference on Spoken Language Translation, Hong Kong. Association for Computational Linguis- tics. Jan Niehues, Thai Son Nguyen, Eunah Cho, Thanh-Le Ha, Kevin Kilgour, Markus Müller, Matthias Sperbe...

    2019 arXiv

  2. [19]

    InProceedings of the 20th International Conference on Spoken Lan- guage Translation (IWSLT 2023) , pages 389– 396, Toronto, Canada (in-person and online)

    Towards efficient simultaneous speech translation: CUNI-KIT system for simultane- ous track at IWSLT 2023. InProceedings of the 20th International Conference on Spoken Lan- guage Translation (IWSLT 2023) , pages 389– 396, Toronto, Canada (in-person and online). Association for...

    2023

  3. [20]

    In Proceedings of the 19th International Confer- ence on Spoken Language Translation (IWSLT 2022), pages 277–285, Dublin, Ireland (in- person and online)

    CUNI-KIT system for simultaneous speech translation task at IWSLT 2022. In Proceedings of the 19th International Confer- ence on Spoken Language Translation (IWSLT 2022), pages 277–285, Dublin, Ireland (in- person and online). Association for Computa- tional Linguistics. Peter...

    2022

  4. [21]

    In Proceedings of the 40th International Conference on Machine Learn- ing, ICML’23

    Shiftable context: addressing training- inference context mismatch in simultaneous speech translation. In Proceedings of the 40th International Conference on Machine Learn- ing, ICML’23. JMLR.org. Dhevi J. Rajendran, Andrew T. Duchowski, Pi- lar Orero, Juan Martínez, and Pablo...

    2013

  5. [22]

    arXiv preprint arXiv:2210.09754

    Simultaneous translation for unseg- mented input: A sliding window approach. arXiv preprint arXiv:2210.09754. Hassan Shavarani, Maryam Siahbani, Ramtin Mehdizadeh Seraj, and Anoop Sarkar

    arXiv

  6. [24]

    arXiv preprint arXiv:2402.01172

    Streaming sequence transduction through dynamic compression. arXiv preprint arXiv:2402.01172. Yun Tang, Anna Sun, Hirofumi Inaguma, Xinyue Chen, Ning Dong, Xutai Ma, Paden Tomasello, and Juan Pino. 2023. Hybrid transducer and attention based encoder-decoder modeling for speech...

    2023 arXiv

  7. [25]

    LAMASSU: A Streaming Language- Agnostic Multilingual Speech Recognition and Translation Model Using Neural Transducers. In Proc. INTERSPEECH 2023, pages 57–61. Xiaolin Wang, Andrew Finch, Masao Utiyama, and Eiichiro Sumita. 2016. An efficient and effective online sentence segm...

    2023

  8. [26]

    In 2008 IEEE Spoken Language Technology Workshop, pages 233–236

    Simultaneous machine translation of german lectures into english: Investigating re- search challenges for the future. In 2008 IEEE Spoken Language Technology Workshop, pages 233–236. Krzysztof Wołk and Krzysztof Marasek. 2014. Real-time statistical speech translation. In New P...

    2008

  9. [27]

    arXiv preprint arXiv:1907.12984

    Dutongchuan: Context-aware transla- tion model for simultaneous interpreting. arXiv preprint arXiv:1907.12984. Jian Xue, Peidong Wang, Jinyu Li, Matt Post, and Yashesh Gaur. 2022. Large-Scale Stream- ing End-to-End Speech Translation with Neural Transducers. In Proc. Interspee...

    1907 arXiv

  10. [28]

    In 2023 IEEE Automatic Speech Recog- nition and Understanding Workshop (ASRU) , pages 1–7

    A weakly-supervised streaming multilin- gual speech model with truly zero-shot capabil- ity. In 2023 IEEE Automatic Speech Recog- nition and Understanding Workshop (ASRU) , pages 1–7. Brian Yan, Jiatong Shi, Soumi Maiti, William Chen, Xinjian Li, Yifan Peng, Siddhant Arora, an...

    2023 arXiv

  11. [30]

    Association for Computational Linguistics

    In Proceedings of the Third Workshop on Automatic Simultaneous Translation , pages 25–33, Online. Association for Computational Linguistics. Biao Zhang, Ivan Titov, Barry Haddow, and Rico Sennrich. 2021. Beyond sentence-level end-to- end speech translation: Context helps. In P...

    2021

  12. [31]

    speech” already in- cludes the results for “speech-to-text

    StreamSpeech: Simultaneous speech- to-speech translation with multi-task learning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages 8964–8986, Bangkok, Thailand. Association for Computa- tional Linguisti...

    2008 arXiv

  13. [1901]

    Francisco Casacuberta, David Llorens, Carlos Martinez, Sirko Molau, Francisco Nevado, Her- mann Ney, Moisés Pastor, David Pico, Al- berto Sanchis, Enrique Vidal, and Juan M

    Curran Associates, Inc. Francisco Casacuberta, David Llorens, Carlos Martinez, Sirko Molau, Francisco Nevado, Her- mann Ney, Moisés Pastor, David Pico, Al- berto Sanchis, Enrique Vidal, and Juan M. Vilar. 2001. Speech-to-speech translation based on finite-state transducers. In...

    2001 arXiv

  14. [2007]

    Machine translation, 21:209–252

    Simultaneous translation of lectures and speeches. Machine translation, 21:209–252. Tomoki Fujita, Graham Neubig, Sakriani Sakti, Tomoki Toda, and Satoshi Nakamura. 2013. Simple, lexicalized choice of translation timing for simultaneous speech translation. In Proc. Interspeech...

    2013

  15. [2008]

    In Proceedings of the 5th Interna- tional Workshop on Spoken Language Transla- tion: Papers, pages 174–181, Waikiki, Hawaii

    Simultaneous German-English lecture translation. In Proceedings of the 5th Interna- tional Workshop on Spoken Language Transla- tion: Papers, pages 174–181, Waikiki, Hawaii. Phillip A Laplante. 1992. Real-time systems de- sign and analysis: an engineer’s handbook . IEEE press....

    1992 arXiv

  16. [2015]

    In Proceedings of the 12th International Workshop on Spoken Language Translation: Papers, pages 217–224, Da Nang, Vietnam

    Learning segmentations that balance latency versus quality in spoken language trans- lation. In Proceedings of the 12th International Workshop on Spoken Language Translation: Papers, pages 217–224, Da Nang, Vietnam. Hiroaki Shimizu, Graham Neubig, Sakriani Sakti, Tomoki Toda, ...

    2013 arXiv

  17. [2016]

    In Proceedings of the 2016 Conference of the North American Chapter of the Associ- ation for Computational Linguistics: Demon- strations, pages 82–86, San Diego, California

    Lecture translator - speech transla- tion framework for simultaneous lecture trans- lation. In Proceedings of the 2016 Conference of the North American Chapter of the Associ- ation for Computational Linguistics: Demon- strations, pages 82–86, San Diego, California. Association...

    2016

  18. [2019]

    In 2019 IEEE Au- tomatic Speech Recognition and Understanding Workshop (ASRU), pages 792–799

    A comparative study on end-to-end speech to text translation. In 2019 IEEE Au- tomatic Speech Recognition and Understanding Workshop (ASRU), pages 792–799. IEEE. Parnia Bahar, Patrick Wilken, Tamer Alkhouli, Andreas Guta, Pavel Golik, Evgeny Matusov, and Christian Herold. 2020...

    2019

  19. [2020]

    In Proceedings of the 17th International Conference on Spoken Language Translation, pages 62–68, Online

    End-to-end simultaneous translation system for IWSLT2020 using modality agnos- tic meta-learning. In Proceedings of the 17th International Conference on Spoken Language Translation, pages 62–68, Online. Association for Computational Linguistics. W. Ronny Huang, Shuo-Yiin Chang...

    2022 arXiv

  20. [2021]

    In Findings of the As- sociation for Computational Linguistics: ACL- IJCNLP 2021, pages 2461–2474, Online

    RealTranS: End-to-end simultaneous speech translation with convolutional weighted- shrinking transformer. In Findings of the As- sociation for Computational Linguistics: ACL- IJCNLP 2021, pages 2461–2474, Online. Asso- ciation for Computational Linguistics. Xingshan Zeng, Peng...

    2021

  21. [2022]

    In Proceedings of the 19th International Conference on Spoken Lan- guage Translation (IWSLT 2022) , pages 177– 189, Dublin, Ireland (in-person and online)

    Efficient yet competitive speech transla- tion: FBK@IWSLT2022. In Proceedings of the 19th International Conference on Spoken Lan- guage Translation (IWSLT 2022) , pages 177– 189, Dublin, Ireland (in-person and online). As- sociation for Computational Linguistics. Marco Gaido, ...

    2022 arXiv

  22. [2023]

    arXiv preprint arXiv:2312.05187

    Seamless: Multilingual expressive and streaming speech translation. arXiv preprint arXiv:2312.05187. Luisa Bentivogli, Mauro Cettolo, Marco Gaido, Alina Karakanta, Alberto Martinelli, Matteo Negri, and Marco Turchi. 2021. Cascade ver- sus direct speech translation: Do the diff...

    2021 arXiv

  23. [2024]

    Transactions of the Association for Computational Linguistics, 12:1104–1121

    Segmentation-Free Streaming Machine Translation. Transactions of the Association for Computational Linguistics, 12:1104–1121. Javier Iranzo-Sánchez, Javier Jorge, Pau Baquero- Arnal, Joan Albert Silvestre-Cerdà, Adrià Giménez, Jorge Civera, Albert Sanchis, and Alfons Juan. 202...

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.