Pith. sign in

REVIEW 4 major objections 5 minor 46 references

PARCO: Phoneme-Augmented Robust Contextual ASR via Contrastive Entity Disambiguation

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read ASR can resist thousands of phonetically similar distractors by disambiguating whole entities at the first token and filtering candidates by phoneme similarity.

desk verdict A competent contextual biasing paper whose strong results are partly self-confirming: both the method and the evaluation rely on phoneme edit distance, so external validity is unproven. read the letter →

arxiv 2509.04357 v1 pith:GIZ5PMKY submitted 2025-09-04 cs.CL cs.AIcs.LGcs.SD

Jiajun He , Naoki Sawada , Koichi Miyazaki , Tomoki Toda This is my paper · ORCID
classification cs.CLcs.AIcs.LGcs.SD
keywords contextualASRphoneme-awareencodingcontrastiveentitydisambiguationhomophoneentity-levelsupervisionhierarchicalfilteringmulti-tokennamedentitiesout-of-domainrobustness
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 claims that context-aware speech recognition can be made robust to homophone-like named entities by explicitly training the decoder to distinguish phonetically similar candidates and by supervising entity retrieval at the level of the whole span, not individual tokens. The proposed method, PARCO, combines a contrastive entity-disambiguation loss, entity-level labeling, and inference-time hierarchical filtering based on phoneme edit distance and confidence gating. If the claim holds, ASR systems can use large biasing lists containing thousands of confusable names without the entity error rates exploding, and they can keep that robustness on out-of-domain speech. This matters because current contextual biasing methods collapse when many acoustically similar options compete, and they often fragment multi-token names.

What carries the argument

The load-bearing components are the contrastive entity-disambiguation (CED) loss, which draws the decoder hidden state near the correct entity's phoneme-enriched representation and repels it from edit-distance-based hard negative entities; entity-level supervision, which labels only the first token of a biased entity with its list index so the model retrieves the whole span at one decoding step; and hierarchical entity filtering (HEF), an inference-only strategy that composes a per-step candidate set from the most-attended entity's phoneme-similar neighbors and gates entity selection on a confidence threshold.

What would settle it

Run PARCO on utterances where the distractors are acoustically confusable under noise but have large phoneme edit distances (e.g., same-sounding names that differ by tone, stress, or dialectal pronunciation), or where the correct entity is deliberately omitted from the biasing list; if entity error rates no longer improve over the no-CED/no-HEF baseline, the claimed mechanism is specific to phoneme-edit-distance distractors and list-present entities.

Watch

Extended reading notes

Core claim

PARCO's central claim is that a phoneme-augmented contextual biasing model can be trained so that the decoder's hidden representation at each step is pushed toward the correct whole entity and away from phonetically similar hard negatives, using an InfoNCE-style contrastive loss. During decoding, an entity-level supervision signal marks only the first token of each entity as the retrieval point, forcing the context attention to select the complete entity span from the biasing list in one step rather than piecewise. At inference, a hierarchical filtering step first finds the most-attended entity, retrieves its top-K phoneme-edit-distance neighbors, and then applies a confidence gate so that l

Load-bearing premise

Phoneme-level edit distance ranks the biasing-list entries that truly compete acoustically with the correct entity in the encoder's representation, and the correct entity is always present in the inference biasing list.

Editorial extensions

If this is right

  • Contextual biasing lists containing thousands of phonetically similar entries become usable without unacceptable entity error inflation, because hard-negative contrastive training sharpens the attention landscape.
  • Multi-token entities such as person and place names are transcribed as intact spans instead of fragmented or partially hallucinated pieces, improving downstream applications like video summarization and emotion recognition.
  • The combined phoneme and text encoders plus entity-level supervision transfer to out-of-domain test sets, suggesting the approach does not require per-domain retuning of the biasing mechanism.
  • The HEF confidence gate provides a principled way to decide when the model should fall back to the standard vocabulary, which could be combined with other confidence-based ASR methods.
  • The ablation results imply that each component—phoneme encoder, CED, entity loss, HEF—contributes additively, so the framework can be partially adopted where full training is infeasible.

Reading between the lines

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

  • The phoneme-edit-distance hard-negative selection assumes that confusability in the acoustics matches confusability in the phoneme transcription; a natural extension would be to replace or augment that ranking with distances computed in the encoder's own acoustic embedding space.
  • The method as presented targets attention-based encoder-decoder models; applying the same entity-level contrastive scheme to RNN-T or streaming decoders would test whether the first-token labeling idea generalizes beyond AED architectures.
  • The confidence threshold sigma is a fixed hyperparameter; an adaptive or calibrated threshold could further reduce false positives on out-of-domain audio without sacrificing recall, though the paper does not explore this.
  • Because the correct entity is assumed present in the inference biasing list, a failure mode would be systematic degradation when the spoken name is absent; the HEF gate partly covers this but a deliberate study of missing-entity cases would clarify the boundary.
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 / 5 minor

Summary. The paper proposes PARCO, a contextual biasing framework for attention-based encoder-decoder ASR that adds four components: phoneme-aware entity encoding, a contrastive entity disambiguation (CED) loss, first-token entity-level supervision, and an inference-time hierarchical entity filter (HEF). The method is evaluated on Chinese AISHELL-1 and English DATA2 with varying numbers of distractors, plus out-of-domain THCHS-30 and LibriSpeech, against CBA, CopyNE, and ED-CEC. The paper reports consistent CER/WER and NE-CER/NE-WER gains, with an ablation study attributing the gains to each component and an attention visualization supporting the CED mechanism. The central claim is that PARCO can disambiguate phonetically confusable entities and retrieve entire multi-token entities more reliably than prior contextual biasing methods.

Significance. If the reported gains are robust, PARCO is a practically useful step for homophone-heavy contextual biasing, especially because the out-of-domain NE-error reductions are large (81.7% on THCHS-30, 50.2% on LibriSpeech relative to the Conformer baseline). The paper has clear strengths: the ablation is monotonic and internally consistent, the arithmetic in Tables II and III checks out, and the attention visualization in Fig. 2 directly illustrates the intended CED effect. However, the main empirical protocol is built around phoneme edit distance for training negatives, inference filtering, and test distractors simultaneously, so the central claim of generalizable acoustic disambiguation is not yet established. The absence of repeated runs, error bars, or significance testing further limits the strength of the comparative claims.

major comments (4)
  1. [Sec. IV-C, Eq. (13), Tables II-III] The training negatives, HEF pre-selection, and inference distractors are all constructed with the same phoneme-level edit-distance criterion (pypinyin/g2pE). This means the model is optimized to suppress exactly the candidate distribution on which it is evaluated. The headline NE-error reductions therefore do not demonstrate transfer to naturally occurring ASR confusions (tone variants, coarticulation effects, or near-homophones that edit distance under-weights). Please validate on at least one distractor set that is not generated by phoneme edit distance, or compare against acoustic confusibility estimated from recognizer confusion pairs, and report how often HEF retains the correct entity when the initial attention anchor is wrong.
  2. [Tables II, III, IV; Abstract] The abstract and Section IV-D use 'significantly outperforming,' but no confidence intervals, standard deviations, or significance tests are reported. Every result appears to be a single run. Given that differences of 0.1-0.2 CER/WER are used to support conclusions, please report mean +/- std over at least 3 seeds and a paired significance test (e.g., bootstrap or approximate randomization) for the main comparisons and ablations.
  3. [Sec. III-D, Eq. (10)] The entity loss labels only the first token of an entity and labels the remaining tokens as C0. The claim that this 'explicitly guides the model to retrieve the entire entity from the biasing list at the correct decoding step' is stronger than what Eq. (10) enforces: after the first token, there is no loss term tying continuation tokens to the selected entity, so partial or fragmented decoding is not directly penalized. Please either add a mechanism that enforces span-level copying or provide a diagnostic showing that first-token selection is followed by complete span retrieval (e.g., compare against full-span token labeling and report span-completeness conditioned on first-token decisions).
  4. [Sec. III-E, Sec. IV-A] HEF depends on two tuned hyperparameters, K=20 and sigma=0.9, and no sensitivity analysis is given. Since HEF first anchors on the currently most-attended entity and then keeps only its top-K phoneme edit-distance neighbors, an early attention error can exclude the correct entity. Please report sweeps over K and sigma and a robustness experiment in which the correct entity is not in the top-K set; this is directly relevant to the claim that HEF 'retains high recall.'
minor comments (5)
  1. [Eq. (5)] The projection weights WN and bN are labeled similarly to the decoder projection Wn/bn in Eqs. (1) and (8). Use distinct notation to avoid confusion.
  2. [Sec. IV-C] Please specify how inference biasing lists are constructed: the pool from which N=0/100/1000/5000 distractors are drawn, whether the ground-truth entity is always included, and whether distractors are sampled with replacement. This is needed for reproducibility and for interpreting the N-scaling results.
  3. [Sec. IV-D] The phrase 'significantly outperforms' appears throughout without statistical support. I recommend reserving 'significant' for tested comparisons.
  4. [Fig. 2] The attention matrices are difficult to read at the printed size. Enlarging the figure or listing the entity indices as a separate legend would help.
  5. [General] No code or data release is mentioned. Given that the protocol has several non-obvious construction choices (hard-negative selection, distractor generation, HEF), public code would materially strengthen reproducibility.

Circularity Check

0 steps flagged · score 2.0 of 10

Empirical comparison with external benchmarks; only minor non-load-bearing self-citation, no circular derivation.

full rationale

The paper's central claims are test-set measurements (CER/WER, NE-CER/NE-WER) obtained under a fixed evaluation protocol against external baselines. No reported number is derived from a fitted constant or from the model's own output by construction. The HEF and CED components use phoneme-level edit distance both for hard-negative sampling and for inference-time filtering, and the test distractor lists are also constructed with phonemically similar distractors; this creates a potential train/eval distribution alignment that raises external-validity concerns, but it is not a circularity because the metrics are computed from recognized text versus ground-truth text, not from edit-distance ranks. K and sigma are tuned, but the final results are on held-out test sets. The only self-citation appearing in the comparison is ED-CEC [19], used as a baseline; it is not load-bearing for any premise of the method. No uniqueness theorem or ansatz is imported from prior self-citation. Therefore no circular step is exhibited.

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

The empirical claims rest on standard AED/Conformer assumptions plus three method-specific premises (phoneme edit distance approximates acoustic confusability; oracle presence of GT entities; first-token supervision enforces span integrity). Five hyperparameters are set by hand or tuned by comparative experiments. No new physical or symbolic entities are introduced.

free parameters (5)
  • HEF pool size K = 20
    Section IV-A: 'based on comparative experiments... ultimately set to 20'; controls how many phonetically similar entities survive filtering.
  • HEF confidence threshold sigma = 0.9
    Section IV-A: tuned by comparative experiments; gates whether biasing is applied at each decoding step.
  • CED temperature tau = 0.1
    Section IV-A: set by hand; controls sharpness of the InfoNCE distribution in Eq. 9.
  • CTC loss weight lambda = 0.7
    Section IV-A: set by hand; balances ASR and CTC terms in Eq. 11.
  • Hard negatives per ground-truth entity = 1-3
    Section IV-C: chosen by hand; the number of negatives in Eq. 9's denominator.
assumptions (4)
  • domain assumption Phoneme-level edit distance over pypinyin and g2pE transcriptions ranks biasing-list entries by real acoustic confusability
    HEF pre-selection (Section III-E) and CED hard-negative retrieval (Section IV-C) both depend on this proxy; the paper never validates it against an acoustic similarity measure.
  • domain assumption The correct entity is always present in the inference biasing list
    Section IV-C constructs lists by 'combining GT entities with a set of phonemically similar distractors'; all NE-CER/NE-WER numbers therefore measure selection among candidates, not detection.
  • ad hoc to paper Labeling only the first token of an entity is sufficient to enforce whole-span entity retrieval
    Section III-D asserts this directly ('guides the model to retrieve the entire entity from the biasing list'), but no mechanism forces the decoder to copy the remaining tokens from the entity representation; the ASR loss must carry that.
  • domain assumption The AED CTC/attention Conformer backbone and its token vocabulary are an adequate base for the biasing study
    Section II fixes the ASR framework; all conclusions are conditional on this backbone and the pypinyin/g2pE phoneme inventories.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PARCO: Phoneme-Augmented Robust Contextual ASR via Contrastive Entity Disambiguation." pith.science (2026). https://pith.science/paper/GIZ5PMKY

@misc{pith2026250904357,
  author       = {Pith},
  title        = {Pith review of: PARCO: Phoneme-Augmented Robust Contextual ASR via Contrastive Entity Disambiguation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GIZ5PMKY}},
  note         = {Machine review of arXiv:2509.04357}
}
read the original abstract

Automatic speech recognition (ASR) systems struggle with domain-specific named entities, especially homophones. Contextual ASR improves recognition but often fails to capture fine-grained phoneme variations due to limited entity diversity. Moreover, prior methods treat entities as independent tokens, leading to incomplete multi-token biasing. To address these issues, we propose Phoneme-Augmented Robust Contextual ASR via COntrastive entity disambiguation (PARCO), which integrates phoneme-aware encoding, contrastive entity disambiguation, entity-level supervision, and hierarchical entity filtering. These components enhance phonetic discrimination, ensure complete entity retrieval, and reduce false positives under uncertainty. Experiments show that PARCO achieves CER of 4.22% on Chinese AISHELL-1 and WER of 11.14% on English DATA2 under 1,000 distractors, significantly outperforming baselines. PARCO also demonstrates robust gains on out-of-domain datasets like THCHS-30 and LibriSpeech.

Figures

Figures reproduced from arXiv: 2509.04357 by the authors.

Figure 1
Figure 1. Overall architecture of the proposed PARCO model. The ASR backbone is based on a Conformer encoder-decoder architecture. A biasing list containing [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Comparison of attention visualization with and without CED loss. The [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 45 canonical work pages

  1. [1]

    Automatic speech recognition using advanced deep learning approaches: A survey,

    H. Kheddar, M. Hemis, and Y . Himeur, “Automatic speech recognition using advanced deep learning approaches: A survey,” Information Fu- sion, pp. 102 422–102 441, 2024

    work page 2024

  2. [2]

    CIF-T: A novel CIF-based transducer architecture for automatic speech recognition,

    T.-H. Zhang, D. Zhou, G. Zhong, J. Zhou, and B. Li, “CIF-T: A novel CIF-based transducer architecture for automatic speech recognition,” in Proc. ICASSP, 2024, pp. 10 531–10 535

    work page 2024

  3. [3]

    Dynamic encoder size based on data-driven layer-wise pruning for speech recognition,

    J. Xu, W. Zhou, Z. Yang, E. Beck, and R. Schl ¨uter, “Dynamic encoder size based on data-driven layer-wise pruning for speech recognition,” in Proc. Interspeech, 2024, pp. 4563–4567

    work page 2024

  4. [4]

    Instant one-shot word-learning for context-specific neural sequence-to-sequence speech recognition,

    C. Huber, J. Hussain, S. St ¨uker, and A. Waibel, “Instant one-shot word-learning for context-specific neural sequence-to-sequence speech recognition,” in Proc. ASRU, 2021, pp. 1–7

    work page 2021

  5. [5]

    Multi-modal video summarization based on two-stage fusion of audio, visual, and recognized text information,

    Z. Yang, J. He, and T. Toda, “Multi-modal video summarization based on two-stage fusion of audio, visual, and recognized text information,” in Proc. APSIPA ASC , 2024, pp. 1–6

    work page 2024

  6. [6]

    2DP-2MRC: 2-dimensional pointer-based machine reading comprehension method for multimodal moment retrieval,

    J. He and T. Toda, “2DP-2MRC: 2-dimensional pointer-based machine reading comprehension method for multimodal moment retrieval,” in Proc. Interspeech, 2024, pp. 5073–5077

    work page 2024

  7. [7]

    MF-AED-AEC: Speech emotion recognition by leveraging multimodal fusion, ASR error detection, and ASR error correction,

    J. He, X. Shi, X. Li, and T. Toda, “MF-AED-AEC: Speech emotion recognition by leveraging multimodal fusion, ASR error detection, and ASR error correction,” in Proc. ICASSP, 2024, pp. 11 066–11 070

    work page 2024

  8. [8]

    Speech Emotion Recognition with ASR Transcripts: A Comprehensive Study on Word Error Rate and Fusion Techniques

    Y . Li, P. Bell, and C. Lai, “Speech emotion recognition with ASR transcripts: A comprehensive study on word error rate and fusion techniques,” arXiv:2406.08353, 2024

    work page Pith review arXiv 2024

Show all 46 references
  1. [9]

    A study on multimodal fusion and layer adapter in emotion recognition,

    X. Shi, Y . Gao, J. He, J. Mi, X. Li, and T. Toda, “A study on multimodal fusion and layer adapter in emotion recognition,” in Proc. APSIPA ASC, 2024, pp. 1–6

    2024

  2. [10]

    Semi-supervised multimodal emotion recognition with consensus decision-making and label correction,

    J. Tian, D. Hu, X. Shi, J. He, X. Li, Y . Gao, T. Toda, X. Xu, and X. Hu, “Semi-supervised multimodal emotion recognition with consensus decision-making and label correction,” in Proc. MRAC, 2023, pp. 67–73

    2023

  3. [11]

    Two-stage framework for robust speech emotion recognition using target speaker extraction in human speech noise conditions,

    J. Mi, X. Shi, D. Ma, J. He, T. Fujimura, and T. Toda, “Two-stage framework for robust speech emotion recognition using target speaker extraction in human speech noise conditions,” in Proc. APSIPA ASC , 2024

    2024

  4. [12]

    GIA-MIC: Multimodal emotion recog- nition with gated interactive attention and modality-invariant learning constraints,

    J. He, J. Mi, and T. Toda, “GIA-MIC: Multimodal emotion recog- nition with gated interactive attention and modality-invariant learning constraints,” in Proc. Interspeech, 2025, pp. 1–5

    2025

  5. [13]

    Tree-constrained pointer generator with graph neural net- work encodings for contextual speech recognition,

    G. Sun et al., “Tree-constrained pointer generator with graph neural net- work encodings for contextual speech recognition,” in Proc. Interspeech, 2022, pp. 2043–2047

    2022

  6. [14]

    MaLa-ASR: Multimedia-assisted LLM-based ASR,

    G. Yang, Z. Ma, F. Yu, Z. Gao, S. Zhang, and X. Chen, “MaLa-ASR: Multimedia-assisted LLM-based ASR,” in Proc. Interspeech, 2024, pp. 2405–2409

    2024

  7. [15]

    Improving neural biasing for contextual speech recognition by early context injec- tion and text perturbation,

    R. Huang, M. Yarmohammadi, S. Khudanpur, and D. Povey, “Improving neural biasing for contextual speech recognition by early context injec- tion and text perturbation,” in Proc. Interspeech, 2024, pp. 752–756

    2024

  8. [16]

    Contextual speech recognition in end-to-end neural network systems using beam search

    I. Williams, A. Kannan, P. S. Aleksic, D. Rybach, and T. N. Sainath, “Contextual speech recognition in end-to-end neural network systems using beam search.” in Proc. Interspeech, 2018, pp. 2227–2231

    2018

  9. [17]

    Deep shallow fusion for RNN-T personalization,

    D. Le, G. Keren, J. Chan, J. Mahadeokar, C. Fuegen, and M. L. Seltzer, “Deep shallow fusion for RNN-T personalization,” in Proc. SLT, 2021, pp. 251–257

    2021

  10. [18]

    Factorized neural transducer for efficient language model adaptation,

    X. Chen, Z. Meng, S. Parthasarathy, and J. Li, “Factorized neural transducer for efficient language model adaptation,” in Proc. ICASSP , 2022, pp. 8132–8136

    2022

  11. [19]

    ED-CEC: Improving rare word recognition using ASR postprocessing based on error detection and context-aware error correction,

    J. He, Z. Yang, and T. Toda, “ED-CEC: Improving rare word recognition using ASR postprocessing based on error detection and context-aware error correction,” in Proc. ASRU, 2023, pp. 1–6

    2023

  12. [20]

    Crossmodal ASR error correction with discrete speech units,

    Y . Li, P. Chen, P. Bell, and C. Lai, “Crossmodal ASR error correction with discrete speech units,” arXiv:2405.16677, 2024

    2024 arXiv

  13. [21]

    Enhancing recognition of rare words in ASR through error detection and context-aware error correction,

    J. He, Z. Yang, and T. Toda, “Enhancing recognition of rare words in ASR through error detection and context-aware error correction,” IEICE Tech. Rep., vol. 123, no. 292, pp. 13–18, 2023

    2023

  14. [22]

    PMF-CEC: Phoneme-augmented multimodal fusion for context-aware asr error correction with error-specific selective decod- ing,

    J. He and T. Toda, “PMF-CEC: Phoneme-augmented multimodal fusion for context-aware asr error correction with error-specific selective decod- ing,” IEEE Transactions on Audio, Speech and Language Processing , vol. 33, pp. 2402–2417, 2025

    2025

  15. [23]

    Deep context: end-to-end contextual speech recognition,

    G. Pundak, T. N. Sainath, R. Prabhavalkar, A. Kannan, and D. Zhao, “Deep context: end-to-end contextual speech recognition,” in Proc. SLT, 2018, pp. 418–425

    2018

  16. [24]

    Improving end-to-end contextual speech recognition with fine-grained contextual knowledge selection,

    M. Han, L. Dong, Z. Liang, M. Cai, S. Zhou, Z. Ma, and B. Xu, “Improving end-to-end contextual speech recognition with fine-grained contextual knowledge selection,” inProc. ICASSP, 2022, pp. 8532–8536

    2022

  17. [25]

    Internal language model estimation through explicit context vector learning for attention- based encoder-decoder ASR,

    Y . Liu, R. Ma, H. Xu, Y . He, Z. Ma, and W. Zhang, “Internal language model estimation through explicit context vector learning for attention- based encoder-decoder ASR,” in Proc. Interspeech , 2022, pp. 1666– 1670

    2022

  18. [26]

    End-to-end contextual ASR based on posterior distribution adaptation for hybrid CTC/attention system,

    Z. Zhang and P. Zhou, “End-to-end contextual ASR based on posterior distribution adaptation for hybrid CTC/attention system,” arXiv:2202.09003, 2022

    2022 arXiv

  19. [27]

    NAM+: Towards scalable end-to-end contextual biasing for adaptive ASR,

    T. Munkhdalai, Z. Wu, G. Pundak, K. C. Sim, J. Li, P. Rondon, and T. N. Sainath, “NAM+: Towards scalable end-to-end contextual biasing for adaptive ASR,” in Proc. SLT, 2023, pp. 190–196

    2023

  20. [28]

    CopyNE: Better contextual ASR by copying named entities,

    S. Zhou, Z. Li, Y . Hong, M. Zhang, Z. Wang, and B. Huai, “CopyNE: Better contextual ASR by copying named entities,” in Proc. ACL, 2024, pp. 2675–2686

    2024

  21. [29]

    Improving contextual asr with enhanced phrase-level representation based on mctc loss,

    M. Fang, T. Wei, K. Guo, Z. Zhuang, Y . Shi, N. Cheng, S. Wang, and J. Xiao, “Improving contextual asr with enhanced phrase-level representation based on mctc loss,” in Proc. ICASSP, 2025, pp. 1–5

    2025

  22. [30]

    Token-level contextual network with ladder-shaped attention for end-to-end asr,

    M. Fang, K. Guo, T. Wei, Z. Zhuang, Y . Shi, N. Cheng, S. Wang, and J. Xiao, “Token-level contextual network with ladder-shaped attention for end-to-end asr,” in Proc. ICASSP, 2025, pp. 1–5

    2025

  23. [31]

    Phoebe: Pronunciation-aware contextualization for end-to-end speech recogni- tion,

    A. Bruguier, R. Prabhavalkar, G. Pundak, and T. N. Sainath, “Phoebe: Pronunciation-aware contextualization for end-to-end speech recogni- tion,” in Proc. ICASSP, 2019, pp. 6171–6175

    2019

  24. [32]

    Joint grapheme and phoneme embeddings for contextual end-to-end ASR

    Z. Chen, M. Jain, Y . Wang, M. L. Seltzer, and C. Fuegen, “Joint grapheme and phoneme embeddings for contextual end-to-end ASR.” in Proc. Interspeech, 2019, pp. 3490–3494

    2019

  25. [33]

    PROCTER: Pronunciation- aware contextual adapter for personalized speech recognition in neural transducers,

    R. Pandey, R. Ren, Q. Luo, J. Liu, A. Rastrow, A. Gandhe, D. Filimonov, G. Strimel, A. Stolcke, and I. Bulyko, “PROCTER: Pronunciation- aware contextual adapter for personalized speech recognition in neural transducers,” in Prco. ICASSP, 2023, pp. 1–5

    2023

  26. [34]

    Phoneme-aware encoding for prefix-tree-based contextual ASR,

    H. Futami, E. Tsunoo, Y . Kashiwagi, H. Ogawa, S. Arora, and S. Watan- abe, “Phoneme-aware encoding for prefix-tree-based contextual ASR,” in Proc. ICASSP, 2024, pp. 10 641–10 645

    2024

  27. [35]

    Joint CTC-attention based end- to-end speech recognition using multi-task learning,

    S. Kim, T. Hori, and S. Watanabe, “Joint CTC-attention based end- to-end speech recognition using multi-task learning,” in Proc. ICASSP, 2017, pp. 4835–4839

    2017

  28. [36]

    Self-distillation for improving CTC-transformer-based ASR systems

    T. Moriya, T. Ochiai, S. Karita, H. Sato, T. Tanaka, T. Ashihara, R. Masumura, Y . Shinohara, and M. Delcroix, “Self-distillation for improving CTC-transformer-based ASR systems.” in Proc. Interspeech, 2020, pp. 546–550

    2020

  29. [37]

    Context-aware transformer transducer for speech recognition,

    F.-J. Chang, J. Liu, M. Radfar, A. Mouchtaris, M. Omologo, A. Rastrow, and S. Kunzmann, “Context-aware transformer transducer for speech recognition,” in Proc. ASRU, 2021, pp. 503–510

    2021

  30. [38]

    Improving ASR contextual biasing with guided attention,

    J. Tang, K. Kim, S. Shon, F. Wu, and P. Sridhar, “Improving ASR contextual biasing with guided attention,” in Proc. ICASSP , 2024, pp. 12 096–12 100

    2024

  31. [39]

    Contex- tualized automatic speech recognition with attention-based bias phrase boosted beam search,

    Y . Sudo, M. Shakeel, Y . Fukumoto, Y . Peng, and S. Watanabe, “Contex- tualized automatic speech recognition with attention-based bias phrase boosted beam search,” in Proc. ICASSP, 2024, pp. 10 896–10 900

    2024

  32. [40]

    Conformer: Convolution- augmented transformer for speech recognition,

    A. Gulati, J. Qin, C.-C. Chiu, N. Parmar, Y . Zhang, J. Yu, W. Han, S. Wang, Z. Zhang, Y . Wu, and R. Pang, “Conformer: Convolution- augmented transformer for speech recognition,” in Proc. Interspeech , 2020, pp. 5036–5040

    2020

  33. [41]

    AISHELL-1: An open- source mandarin speech corpus and a speech recognition baseline,

    H. Bu, J. Du, X. Na, B. Wu, and H. Zheng, “AISHELL-1: An open- source mandarin speech corpus and a speech recognition baseline,” in Proc. O-COCOSDA, 2017, pp. 1–5

    2017

  34. [42]

    End-to-end named entity recognition from english speech,

    H. Yadav, S. Ghosh, Y . Yu, and R. R. Shah, “End-to-end named entity recognition from english speech,” in Proc. Interspeech, 2020, pp. 4268– 4272

    2020

  35. [43]

    THCHS-30: A free Chinese speech corpus,

    D. Wang and X. Zhang, “THCHS-30: A free Chinese speech corpus,” arXiv:1512.01882, 2015

    2015 arXiv

  36. [44]

    LibriSpeech: An ASR corpus based on public domain audio books,

    V . Panayotov, G. Chen, D. Povey, and S. Khudanpur, “LibriSpeech: An ASR corpus based on public domain audio books,” in Proc. ICASSP , 2015, pp. 5206–5210

    2015

  37. [45]

    CIF-based collaborative de- coding for end-to-end contextual speech recognition,

    M. Han, L. Dong, S. Zhou, and B. Xu, “CIF-based collaborative de- coding for end-to-end contextual speech recognition,” in Proc. ICASSP, 2021, pp. 6528–6532

    2021

  38. [46]

    AISHELL- NER: Named entity recognition from Chinese speech,

    B. Chen, G. Xu, X. Wang, P. Xie, M. Zhang, and F. Huang, “AISHELL- NER: Named entity recognition from Chinese speech,” in Proc. ICASSP, 2022, pp. 8352–8356

    2022

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.