Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Locate-and-Focus: Enhancing Terminology Translation in Speech Language Models

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Direct speech translation misses terms; Locate-and-Focus finds them first, then translates them.

desk verdict A genuinely useful recipe for clip-level terminology retrieval in speech translation, with the caveat that the reported gains assume the dictionary already contains every term. read the letter →

arxiv 2507.18263 v1 pith:FXQ6C7IT submitted 2025-07-24 cs.CL cs.AI

Suhang Wu , Jialong Tang , Chengyi Yang , Pei Zhang , Baosong Yang , Junhui Li , Junfeng Yao , Min Zhang
show 1 more author
Jinsong Su
This is my paper · ORCID
classification cs.CLcs.AI
keywords terminologytranslationspeechlanguagemodelsretrieval-augmentedcliplocalizationslidingwindowretrievalaudioreplacementtermsuccessrate
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

Direct speech translation often mangles names, places, and domain terms, and existing knowledge-injection methods drown the model in irrelevant context or mismatched speakers. Locate-and-Focus claims to fix this in two moves: it first uses a sliding-window speech retriever to find the exact audio clip where the term is spoken, then feeds the model a knowledge triplet in which that located clip replaces the dictionary's synthetic audio, so the utterance and the knowledge share acoustics. A special tag is inserted before the term's translation during training, acting as a self-reminder to consult the triplet. On CoVoST2, MuST-C, and MSLT, the method raises Term Success Rate from roughly 24–53% to 65–77% in the end-to-end setting while keeping BLEU comparable to translation-training baselines.

What carries the argument

The load-bearing object is the translation knowledge triplet $K=(x,c,y)$—term transcript, term speech clip, term translation—and the operation that reshapes it. Sliding Retrieval encodes the clip $c$ and utterance $u$ with the same speech encoder, then takes $\max_i \text{Cosine}(\text{MaxPool}(z_c), \text{MaxPool}(z_u^i))$ over windows $z_u^i$ of length $|c|$, which both scores the term's presence and returns the located subsequence $s$. Audio Replacement swaps $c$ for $s$ to form $K'=(x,s,y)$, giving utterance and knowledge identical acoustics; Tag Cue trains the model to predict <Term> before each terminology translation. The contrastive loss on the speech encoder and the next-token loss on the speech LLM (via LoRA) train the two steps sequentially.

What would settle it

Take the end-to-end setup and remove one term from the retrieval pool while keeping the rest; if Term Success Rate does not fall for that term substantially more than for pool-covered terms, the gain is not attributable to retrieval. Alternatively, replace the pool's synthetic audio with a different speaker's pronunciation of the same term but keep the located clip replaced: if the method's advantage over the audio-replacement ablation disappears, the benefit is specifically the acoustic match, not the presence of the transcript and translation.

Watch

Extended reading notes

Core claim

The paper establishes that clip-level localization is the missing step in knowledge-augmented direct speech translation. Given an utterance and a pool of term triplets (transcript, speech clip, translation), a max-pooled cosine similarity between the utterance embedding and each short clip—computed over every sliding window of the clip's length—both identifies the terms present and pinpoints the acoustic span containing them. Substituting that located span for the pool's synthetic clip creates an acoustic anchor shared by the utterance and the injected knowledge, and training the model to emit <Term> before translating the term lets it attend to that knowledge at the right moment. In the oracle setting the method reaches 86–96% Term Success Rate across six dataset–direction pairs; in the end-to-end setting it outperforms both the collect-and-integrate and retrieve-and-demonstrate baselines by large margins (e.g., 65.5 vs. 28.2 and 32.9 TSR on CoVoST2 EN–ZH), with BLEU gains rather than losses on most pairs.

Load-bearing premise

The knowledge base must already contain the term with a correct transcript, a usable audio clip, and a correct translation, and the evaluation pool is built from the very transcripts being tested, so the retriever never faces genuinely out-of-vocabulary terms.

Editorial extensions

If this is right

  • In end-to-end use, Term Success Rate rises to 65–77% across three datasets (from 24–53% at baseline), and the method beats both the collect-and-integrate and retrieve-and-demonstrate baselines on TSR in every tested setting.
  • General translation quality does not suffer: BLEU on the standard CoVoST2 test set (43.48 EN–ZH) is close to translation-training (43.64) and above the other knowledge-based methods.
  • Clip localization is robust across encoder scales: Sliding Retrieval with Whisper-base, -small, and -medium consistently outperforms pooling-based retrieval, reaching about 60% Hits@1 and 85% Hits@10 on CoVoST2 with the medium encoder.
  • Each component contributes: removing Audio Replacement and Tag Cue together drops oracle TSR (e.g., 86.14 vs. 94.09 on MuST-C EN–ZH), and removing Sliding Retrieval degrades end-to-end TSR.
  • Retrieval adds negligible latency (0.217 ms vs. 621.951 ms for translation itself), so the method is compatible with real-time constraints.

Reading between the lines

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

  • Not claimed in the paper: the same locate-then-focus recipe could transfer to other span-level speech tasks such as hotword customization in ASR or keyword spotting, where the acoustic anchor can replace the dictionary audio.
  • This goes beyond the paper: the method's reliance on a complete, acoustically matched knowledge base is a boundary condition—if the pool is built from the test transcripts themselves, in-the-wild gains will shrink whenever a term is missing or pronounced differently; a testable extension is to measure TSR when the pool is contaminated with distractor terms.
  • The top-5 sweet spot suggests an optimizable tradeoff between recall and noise; a dynamic similarity threshold, rather than a fixed $k$, could further improve end-to-end TSR.
  • The <Term> tag effectively teaches the model when to consult the retrieved knowledge—an attention-control mechanism that could be reused in other retrieval-augmented generation settings.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes Locate-and-Focus, a method for terminology translation in speech-to-text LLMs. The method first runs a sliding-window audio retriever over the input utterance to locate speech clips that contain terminology and to retrieve matching knowledge triplets (term transcript, TTS audio, translation) from a knowledge base. It then feeds the utterance together with the retrieved triplets to a speech LLM, replacing the TTS audio in each triplet with the located utterance clip (Audio Replacement) and training the model to emit a <Term> tag before translating a term (Tag Cue). The authors build a terminology-translation dataset from CoVoST2, MuST-C, and MSLT with LLM-assisted extraction and manual verification, covering English-to-Chinese and English-to-German. Experiments in oracle and end-to-end knowledge settings report large Term Success Rate (TSR) gains over SALM and Retrieval-and-Demonstration baselines, with comparable BLEU, and ablations indicate that each component contributes. The paper also reports retrieval Hits@N, a manual evaluation of located clips, inference latency, and general-domain BLEU on standard CoVoST2.

Significance. If the results hold, the paper offers a practical and reasonably novel approach to terminology translation in speech LLMs. The sliding-window clip-level retrieval is a sensible way to localize terminology in the audio stream, and the audio-replacement idea is an elegant mechanism for making dictionary knowledge acoustically consistent with the utterance. The paper is thorough in its ablations, model-size scaling, latency measurements, and manual verification of located clips. The main empirical claim, however, is currently demonstrated under a closed-vocabulary evaluation protocol: the retrieval pool is constructed from the same test transcripts used for evaluation, so every target term is guaranteed to have a correct triplet in the pool. The reported end-to-end TSR is therefore a conditional success rate at dictionary coverage equal to 1.0, and the paper does not quantify how performance degrades when the knowledge base is incomplete or contains incorrect entries. Because the method's central value proposition depends on the knowledge base, this external-validity gap is load-bearing and needs to be addressed with additional experiments.

major comments (4)
  1. [Section 4, Table 1; Section 5.1] The end-to-end evaluation is closed-vocabulary by construction. For MuST-C and MSLT, the retrieval pool P is built from the test split itself; for CoVoST2, from train and test splits. Consequently, every terminology that must be translated is guaranteed to appear in P with a correct transcript, TTS audio, and golden translation at inference time. The reported end-to-end TSR (65.5-77.4 across settings) is a conditional success rate given full KB coverage, not a measure of robustness under incomplete or incorrect knowledge. The Limitations section acknowledges the reliance on predefined terminologies, but no experiment quantifies the degradation when a term is absent from the KB. Please add a held-out-term evaluation or a KB-coverage sweep (e.g., randomly dropping 10%, 25%, 50% of test terms from the pool) and report the resulting TSR and BLEU.
  2. [Section 5.1, Table 2] The ablation claims are not supported by variance estimates or significance tests. For example, in the CoVoST2 English-to-Chinese end-to-end setting, full Locate-and-Focus achieves 65.53 TSR, w/o Audio Replacement 63.49, w/o Tag Cue 63.73, and w/o Replacement and Cue 62.95; the oracle-setting gaps are similarly small (e.g., 90.13 vs 89.67 vs 89.00). With single runs and no error bars, these differences may be within noise. Please report results over multiple random seeds with standard deviations, or at least provide bootstrap confidence intervals and a paired significance test for the key comparisons in Table 2.
  3. [Section 5, Setups] The Oracle Knowledge Setting for the SALM baseline is under-specified. The paper states that SALM is provided with 'the translations of the top 50 most frequent terms,' but it does not state whether term frequency is computed on the training split, the test split, or the retrieval pool. If the frequency list is computed on the test split, SALM's oracle performance reflects test-set statistics and gives the baseline an unfair advantage; if computed on training data, the setup is not truly 'ground truth knowledge without irrelevant noise.' Please specify the exact construction of the frequency list in both the oracle and end-to-end settings.
  4. [Section 3.1, Eq. (1); Appendix A.1] There is an inconsistency in how the located clip s is extracted. Equation (1) identifies the subsequence of length |c| with maximum pooled-cosine similarity, which would make s exactly as long as the KB clip c. Appendix A.1, however, states that 'we focus on the hidden state with the highest similarity' and segment 'the relevant portion,' suggesting a frame-level selection rather than a full-window selection. Because Audio Replacement and the reported clip-localization accuracy in Section 5.2 depend on the exact segmentation rule, please clarify the precise algorithm used to produce s.
minor comments (6)
  1. [Section 5.4, Table 5] It is not stated whether the standard CoVoST2 evaluation is performed with or without retrieval knowledge and whether <Term> tags are stripped before computing BLEU. Please clarify the inference protocol and the post-processing of tag tokens.
  2. [Section 5, Metrics] The Term Success Rate metric is defined only as 'the proportion of terminologies accurately translated within an utterance.' Please specify the matching criterion (exact substring match with the reference translation, normalized match, etc.) and how the <Term> tag is handled in TSR computation, especially for inflected target languages like German.
  3. [Section 4, D.1] The manual annotation process is described as requiring three-expert agreement, but no inter-annotator agreement statistic (e.g., Cohen's kappa) is reported. Please add agreement numbers to support the claim of high annotation quality.
  4. [Appendix B.2, Table 10] The table header and text use 'BELU' instead of 'BLEU'; please fix this typo.
  5. [Appendix D.2] The instruction template for SALM includes an 'Audio: <audio>...</audio>' field even though SALM is described as a text-only term-list method. Please ensure the template matches the actual input format used in the SALM experiments.
  6. [Section 3, last paragraph] The sentence 'we are the first end-to-end terminology translation method that retrieves and fully utilizes multi-modal fined-granularity multi-modal fine-grained knowledge' contains a repeated and malformed phrase ('fined-granularity multi-modal fine-grained'); please rephrase.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation found; the test-derived retrieval pool is a coverage and validity limitation, not a construction-level circularity.

full rationale

This paper is an empirical engineering contribution rather than a formal derivation, and none of its equations reduces to its own inputs. Equation (1) defines a sliding-window cosine similarity for retrieval, Equation (2) defines a contrastive loss for the speech encoder, and Equation (3) defines a next-token prediction loss for the speech LLM; none of these definitions makes the reported TSR, BLEU, or Hits@N values true by construction. The retrieval pool is indeed built from the same transcripts used for evaluation (Section 4, Table 1), so every target terminology is guaranteed to have a correct triplet in the pool, and the authors explicitly acknowledge the reliance on predefined terminologies in the Limitations section. This is a genuine external-validity threat: the end-to-end TSR numbers assume complete dictionary coverage and do not quantify degradation when a term is absent or mispronounced. However, that is not circularity under the stated standard. The reported scores still require the retriever to rank the correct candidate above distractors and the speech LLM to produce the correct translation, so the numbers are not forced by construction. The Oracle Knowledge Setting is explicitly presented as an ideal-conditions upper bound, not as a prediction. There is no load-bearing self-citation, no imported uniqueness theorem, and no ansatz smuggled in via citation. Accordingly, the appropriate finding is no significant circularity, with score 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper is empirical, so the ledger lists the hand-chosen hyperparameters and the domain assumptions on which the pipeline depends. There are no newly postulated physical entities; the <Term> tag is a training artifact, not an independent entity.

free parameters (5)
  • top_k number of retrieved knowledge triplets = 5
    Table 4 shows top-5 outperforms top-1 and top-10 across datasets; this is a hand-selected hyperparameter evaluated on the test sets.
  • sliding window size = length of the term clip |c|
    The window size equals the encoded clip length by design (Section 3.1); this choice assumes the clip spans exactly the terminology.
  • sliding window step size = 1 frame (about 0.02 seconds)
    Step size 1 is chosen to maximize localization granularity; it increases compute but is parallelized.
  • number of negative samples for retriever training = 4
    Section 5 and Appendix A.1 set 4 negatives per example and 3 epochs; chosen for training efficiency, not justified by analysis.
  • edit distance threshold for TTS clip filtering = 3
    Section 4 retains synthesized clips whose ASR transcript is within edit distance 3 of the source term; this threshold is arbitrary but used to filter low-quality clips.
assumptions (4)
  • domain assumption The knowledge base triplets K=(x,c,y) contain correct term text, a TTS audio clip that matches the term, and a correct translation.
    Section 4 builds the pool by LLM extraction, TTS synthesis, and manual review; errors in any component propagate into retrieval and translation.
  • domain assumption Max-pooled cosine similarity over sliding windows of Whisper speech embeddings identifies the audio span of a term.
    Section 3.1, Equation 1; this assumes frame-level embeddings are phonetically aligned and that max pooling suppresses non-term content.
  • domain assumption Replacing the TTS clip with the located utterance clip preserves the triplet's meaning and helps the LLM associate utterance and knowledge.
    Section 3.2 Audio Replacement; if the located clip contains surrounding speech, the association could be noisy.
  • domain assumption Training the model to output a <Term> token before term translations transfers to inference as a reliable cue.
    Section 3.2 Tag Cue; the model must learn to predict the tag at the right time, and the tag must trigger knowledge usage.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Locate-and-Focus: Enhancing Terminology Translation in Speech Language Models." pith.science (2026). https://pith.science/paper/FXQ6C7IT

@misc{pith2026250718263,
  author       = {Pith},
  title        = {Pith review of: Locate-and-Focus: Enhancing Terminology Translation in Speech Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FXQ6C7IT}},
  note         = {Machine review of arXiv:2507.18263}
}
read the original abstract

Direct speech translation (ST) has garnered increasing attention nowadays, yet the accurate translation of terminology within utterances remains a great challenge. In this regard, current studies mainly concentrate on leveraging various translation knowledge into ST models. However, these methods often struggle with interference from irrelevant noise and can not fully utilize the translation knowledge. To address these issues, in this paper, we propose a novel Locate-and-Focus method for terminology translation. It first effectively locates the speech clips containing terminologies within the utterance to construct translation knowledge, minimizing irrelevant information for the ST model. Subsequently, it associates the translation knowledge with the utterance and hypothesis from both audio and textual modalities, allowing the ST model to better focus on translation knowledge during translation. Experimental results across various datasets demonstrate that our method effectively locates terminologies within utterances and enhances the success rate of terminology translation, while maintaining robust general translation performance.

Figures

Figures reproduced from arXiv: 2507.18263 by the authors.

Figure 1
Figure 1. The differences between Locate-and-Focus and the existing paradigms. We use [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of the Locate-and-Focus method, which comprises the speech terminology clip localization and the terminology-focused translation steps. For a given utterance, the first step effectively identifies and locates speech clips within utterances containing the terminology. Subsequently, the second step uses audio replacement to associate the utterance and translation knowledge through their shared speech clip. It… view at source ↗
Figure 3
Figure 3. Comparison of Hits@5 scores for different [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Attention2Probability: Attention-Driven Terminology Probability Estimation for Robust Speech-to-Text System

    cs.CL 2025-08 conditional novelty 6.0 of 10

    A cross-attention term retriever estimates which terminology appears in speech and, when its top-k terms are added to the prompt, improves SLM terminology accuracy by 6-17%.

Reference graph

Works this paper leans on

44 extracted references · 16 canonical work pages · cited by 1 Pith paper

  1. [1]

    Melissa Ailem, Jingshu Liu, and Raheel Qader. 2022. https://aclanthology.org/2022.jeptalnrecital-taln.44 Encouraging neural machine translation to satisfy terminology constraints . In Actes de la 29e Conf \' e rence sur le Traitement Automatique des Langues Naturelles. Volume 1 : conf \' e rence principale, TALN-RECITAL 2022, Avignon, France, June 27 - Ju...

    work page 2022

  2. [2]

    Keyu An, Qian Chen, Chong Deng, Zhihao Du, Changfeng Gao, Zhifu Gao, Yue Gu, Ting He, Hangrui Hu, Kai Hu, Shengpeng Ji, Yabin Li, Zerui Li, Heng Lu, Haoneng Luo, Xiang Lv, Bin Ma, Ziyang Ma, Chongjia Ni, Changhe Song, Jiaqi Shi, Xian Shi, Hao Wang, Wen Wang, Yuxuan Wang, Zhangyu Xiao, Zhijie Yan, Yexin Yang, Bin Zhang, Qinglin Zhang, Shiliang Zhang, Nan Z...

    arXiv doi:10.48550/arxiv.2407.04051 2024

  3. [3]

    Toms Bergmanis and Marcis Pinnis. 2021. https://doi.org/10.18653/V1/2021.EACL-MAIN.271 Facilitating terminology translation with target lemma annotations . In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, EACL 2021, Online, April 19 - 23, 2021 , pages 3105--3111. Association for C...

    doi:10.18653/v1/2021.eacl-main.271 2021

  4. [4]

    Nikolay Bogoychev and Pinzhen Chen. 2023. https://doi.org/10.18653/V1/2023.WMT-1.80 Terminology-aware translation with constrained decoding and large language model prompting . In Proceedings of the Eighth Conference on Machine Translation, WMT 2023, Singapore, December 6-7, 2023 , pages 890--896. Association for Computational Linguistics

    doi:10.18653/v1/2023.wmt-1.80 2023

  5. [5]

    Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert - Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litw...

    2020

  6. [6]

    Roldano Cattoni, Mattia Antonino Di Gangi, Luisa Bentivogli, Matteo Negri, and Marco Turchi. 2021. https://doi.org/10.1016/J.CSL.2020.101155 Must-c: A multilingual corpus for end-to-end speech translation . Comput. Speech Lang., 66:101155

    arXiv 2021

  7. [7]

    Guanhua Chen, Yun Chen, and Victor O. K. Li. 2021. https://doi.org/10.1609/AAAI.V35I14.17496 Lexically constrained neural machine translation with explicit alignment guidance . In Thirty-Fifth AAAI Conference on Artificial Intelligence, AAAI 2021, Thirty-Third Conference on Innovative Applications of Artificial Intelligence, IAAI 2021, The Eleventh Sympos...

    work page doi:10.1609/aaai.v35i14.17496 2021

  8. [9]

    Yunfei Chu, Jin Xu, Qian Yang, Haojie Wei, Xipin Wei, Zhifang Guo, Yichong Leng, Yuanjun Lv, Jinzheng He, Junyang Lin, Chang Zhou, and Jingren Zhou. 2024. https://doi.org/10.48550/ARXIV.2407.10759 Qwen2-audio technical report . CoRR, abs/2407.10759

    arXiv doi:10.48550/arxiv.2407.10759 2024

Show all 44 references
  1. [10]

    Simone Conia, Daniel Lee, Min Li, Umar Farooq Minhas, Saloni Potdar, and Yunyao Li. 2024. https://aclanthology.org/2024.emnlp-main.914 Towards cross-cultural machine translation with retrieval-augmented generation from multilingual knowledge graphs . In Proceedings of the 2024...

    2024

  2. [11]

    Josep Maria Crego, Jungi Kim, Guillaume Klein, Anabel Rebollo, Kathy Yang, Jean Senellart, Egor Akhanov, Patrice Brunelle, Aur \' e lien Coquard, Yongchao Deng, Satoshi Enoue, Chiyo Geiss, Joshua Johanson, Ardas Khalsa, Raoum Khiari, Byeongil Ko, Catherine Kobus, Jean Lorieux,...

    2016 arXiv

  3. [12]

    Georgiana Dinu, Prashant Mathur, Marcello Federico, and Yaser Al - Onaizan. 2019. https://doi.org/10.18653/V1/P19-1294 Training neural machine translation to apply terminology constraints . In Proceedings of the 57th Conference of the Association for Computational Linguistics,...

    2019 doi

  4. [13]

    Zhihao Du, Yuxuan Wang, Qian Chen, Xian Shi, Xiang Lv, Tianyu Zhao, Zhifu Gao, Yexin Yang, Changfeng Gao, Hui Wang, Fan Yu, Huadai Liu, Zhengyan Sheng, Yue Gu, Chong Deng, Wen Wang, Shiliang Zhang, Zhijie Yan, and Jingren Zhou. 2024. https://doi.org/10.48550/ARXIV.2412.10117 C...

    2024 doi arXiv

  5. [14]

    Christian Federmann and William D. Lewis. 2016. https://aclanthology.org/2016.iwslt-1.12 Microsoft speech language translation (MSLT) corpus: The IWSLT 2016 release for english, french and german . In Proceedings of the 13th International Conference on Spoken Language Translat...

    2016

  6. [15]

    Christian Federmann and William D. Lewis. 2017. https://aclanthology.org/2017.mtsummit-papers.6 The microsoft speech language translation (MSLT) corpus for chinese and japanese: Conversational test data for machine translation and speech recognition . In Proceedings of Machine...

    2017

  7. [16]

    Marco Gaido, Yun Tang, Ilia Kulikov, Rongqing Huang, Hongyu Gong, and Hirofumi Inaguma. 2023. https://doi.org/10.1109/ICASSP49357.2023.10094689 Named entity detection and injection for direct speech translation . In IEEE International Conference on Acoustics, Speech and Signal...

    2023

  8. [17]

    Mahendra Gupta, Maitreyee Dutta, and Chandresh Kumar Maurya. 2024. https://doi.org/10.48550/ARXIV.2411.14453 Direct speech-to-speech neural machine translation: A survey . CoRR, abs/2411.14453

    2024 doi arXiv

  9. [18]

    Minglun Han, Linhao Dong, Zhenlin Liang, Meng Cai, Shiyu Zhou, Zejun Ma, and Bo Xu. 2022. https://doi.org/10.1109/ICASSP43922.2022.9747101 Improving end-to-end contextual speech recognition with fine-grained contextual knowledge selection . In IEEE International Conference on ...

    2022

  10. [19]

    Eva Hasler, Adri \` a de Gispert, Gonzalo Iglesias, and Bill Byrne. 2018. https://doi.org/10.18653/V1/N18-2081 Neural machine translation decoding with terminology constraints . In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computat...

    2018 doi

  11. [20]

    Chris Hokamp and Qun Liu. 2017. https://doi.org/10.18653/V1/P17-1141 Lexically constrained decoding for sequence generation using grid beam search . In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, ACL 2017, Vancouver, Canada, July 30...

    2017 doi

  12. [21]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen - Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen - Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. https://openreview.net/forum?id=nZeVKeeFYf9 Lora: Low-rank adaptation of large language models . In The Tenth International Conference on Learning Representat...

    2022

  13. [22]

    Jiliang Hu, Zuchao Li, Ping Wang, Haojun Ai, Lefei Zhang, and Hai Zhao. 2024. https://aclanthology.org/2024.emnlp-main.821 VHASR: A multimodal speech recognition system with vision hotwords . In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Proces...

    2024

  14. [24]

    Hui Jiang, Ziyao Lu, Fandong Meng, Chulun Zhou, Jie Zhou, Degen Huang, and Jinsong Su. 2022. https://doi.org/10.18653/V1/2022.EMNLP-MAIN.367 Towards robust k-nearest-neighbor machine translation . In Proceedings of the 2022 Conference on Empirical Methods in Natural Language P...

    2022 doi

  15. [25]

    Siqi Li, Danni Liu, and Jan Niehues. 2024 a . https://aclanthology.org/2024.emnlp-main.708 Optimizing rare word accuracy in direct speech translation with a retrieval-and-demonstration approach . In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Pr...

    2024

  16. [26]

    Yuang Li, Yinglu Li, Min Zhang, Chang Su, Jiawei Yu, Mengyao Piao, Xiaosong Qiao, Miaomiao Ma, Yanqing Zhao, and Hao Yang. 2024 b . https://aclanthology.org/2024.lrec-main.262 Cb-whisper: Contextual biasing whisper using open-vocabulary keyword-spotting . In Proceedings of the...

    2024

  17. [27]

    Jiarui Liu, Iman Ouzzani, Wenkai Li, Lechen Zhang, Tianyue Ou, Houda Bouamor, Zhijing Jin, and Mona Diab. 2025. http://arxiv.org/abs/2412.18367 Towards global ai inclusivity: A large-scale multilingual terminology dataset (gist)

    2025 arXiv

  18. [28]

    Zhongjian Miao, Xiang Li, Liyan Kang, Wen Zhang, Chulun Zhou, Yidong Chen, Bin Wang, Min Zhang, and Jinsong Su. 2022. https://aclanthology.org/2022.coling-1.468 Towards robust neural machine translation with iterative scheduled data-switch training . In Proceedings of the 29th...

    2022

  19. [29]

    Elise Michon, Josep Maria Crego, and Jean Senellart. 2020. https://doi.org/10.18653/V1/2020.COLING-MAIN.348 Integrating domain terminology into neural machine translation . In Proceedings of the 28th International Conference on Computational Linguistics, COLING 2020, Barcelona...

    2020 doi

  20. [30]

    Sara Papi, Marco Turchi, and Matteo Negri. 2023. https://doi.org/10.21437/INTERSPEECH.2023-170 Alignatt: Using attention-based audio-translation alignments as a guide for simultaneous speech translation . In 24th Annual Conference of the International Speech Communication Asso...

    2023 doi

  21. [31]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei - Jing Zhu. 2002. https://doi.org/10.3115/1073083.1073135 Bleu: a method for automatic evaluation of machine translation . In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, July 6-12, ...

    2002

  22. [32]

    Jing Peng, Yucheng Wang, Yu Xi, Xu Li, Xizhuo Zhang, and Kai Yu. 2024. http://arxiv.org/abs/2410.18908 A survey on speech large language models

    2024

  23. [33]

    Matt Post and David Vilar. 2018. https://doi.org/10.18653/V1/N18-1119 Fast lexically constrained decoding with dynamic beam allocation for neural machine translation . In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Ling...

    2018 doi

  24. [34]

    Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2023. https://proceedings.mlr.press/v202/radford23a.html Robust speech recognition via large-scale weak supervision . In International Conference on Machine Learning, ICML 2023, 23-29 J...

    2023

  25. [35]

    Shangeth Rajaa and Abhinav Tushar. 2024. https://github.com/skit-ai/SpeechLLM SpeechLLM: Multi-Modal LLM for Speech Understanding

    2024

  26. [36]

    Kirill Semenov, Vilém Zouhar, Tom Kocmi, Dongdong Zhang, Wangchunshu Zhou, and Yuchen Eleanor Jiang. 2023. Findings of the wmt 2023 shared task on machine translation with terminologies. In Proceedings of the Eight Conference on Machine Translation (WMT). Association for Compu...

    2023

  27. [37]

    Nivedita Sethiya and Chandresh Kumar Maurya. 2025. https://doi.org/10.1016/J.CSL.2024.101751 End-to-end speech-to-text translation: A survey . Comput. Speech Lang., 90:101751

    2025

  28. [38]

    Xian Shi, Yexin Yang, Zerui Li, Yanni Chen, Zhifu Gao, and Shiliang Zhang. 2024. https://doi.org/10.1109/ICASSP48485.2024.10446106 Seaco-paraformer: A non-autoregressive ASR system with flexible and effective hotword customization ability . In IEEE International Conference on ...

    2024

  29. [39]

    Changhan Wang, Anne Wu, and Juan Miguel Pino. 2020. http://arxiv.org/abs/2007.10310 Covost 2: A massively multilingual speech-to-text translation corpus . CoRR, abs/2007.10310

    2020 arXiv

  30. [40]

    Shuo Wang, Zhixing Tan, and Yang Liu. 2022. https://doi.org/10.18653/V1/2022.ACL-LONG.487 Integrating vectorized lexical constraints for neural machine translation . In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pape...

    2022 doi

  31. [41]

    An Yang, Beichen Zhang, Binyuan Hui, Bofei Gao, Bowen Yu, Chengpeng Li, Dayiheng Liu, Jianhong Tu, Jingren Zhou, Junyang Lin, Keming Lu, Mingfeng Xue, Runji Lin, Tianyu Liu, Xingzhang Ren, and Zhenru Zhang. 2024. https://doi.org/10.48550/ARXIV.2409.12122 Qwen2.5-math technical...

    2024 doi arXiv

  32. [42]

    Yongjing Yin, Jiali Zeng, Yafu Li, Fandong Meng, and Yue Zhang. 2024. https://aclanthology.org/2024.findings-emnlp.866 Lexmatcher: Dictionary-centric data curation for llm-based machine translation . In Findings of the Association for Computational Linguistics: EMNLP 2024, Mia...

    2024

  33. [43]

    Urchade Zaratiana, Nadi Tomeh, Pierre Holat, and Thierry Charnois. 2024. https://doi.org/10.18653/V1/2024.NAACL-LONG.300 Gliner: Generalist model for named entity recognition using bidirectional transformer . In Proceedings of the 2024 Conference of the North American Chapter ...

    2024 doi

  34. [44]

    Huaao Zhang, Qiang Wang, Bo Qin, Zelin Shi, Haibo Wang, and Ming Chen. 2023. https://doi.org/10.18653/V1/2023.ACL-LONG.332 Understanding and improving the robustness of terminology constraints in neural machine translation . In Proceedings of the 61st Annual Meeting of the Ass...

    2023 doi

  35. [45]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before...

  36. [46]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.