Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Few-shot Personalization via In-Context Learning for Speech Emotion Recognition based on Speech-Language Model

T0 review · 4 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a speech-language model meta-trained for in-context learning can personalize speech emotion recognition to unseen speakers by conditioning on a few enrollment utterance-label pairs, improving unweighted accuracy from 0

desk verdict A solid new application of MetaICL to SER personalization, but the private acted dataset leaves the main result under-validated. read the letter →

arxiv 2509.08344 v1 pith:CGGJBI7B submitted 2025-09-10 eess.AS

Mana Ihori , Taiga Yamane , Naotaka Kawata , Naoki Makishima , Tomohiro Tanaka , Satoshi Suzuki , Shota Orihashi , Ryo Masumura This is my paper
classification eess.AS
keywords speechemotionrecognitionin-contextlearningpersonalizationspeech-languagemodelmeta-trainingfew-shotunseenspeakersspeakeradaptation
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

The paper proposes that speech emotion recognition can be personalized to a new speaker by feeding a few labeled emotional utterances from that speaker into a speech-language model as in-context examples, with no parameter updates. The authors meta-train the model on 800 Japanese speakers so it learns to infer a target utterance's emotion from enrollment utterance-label pairs of the same speaker. On their new dataset, accuracy rises from 0.675 with no examples to 0.757 with seven, beating enrollment-based classifier baselines (best 0.626). If this works in practice, it would let SER systems adapt to a user's emotional expression with a handful of clips, without retraining.

What carries the argument

The carry mechanism is the speech-LM architecture paired with MetaICL training. A frozen transformer speech encoder feeds a Q-Former (a fixed-length query transformer) whose output is concatenated with a text instruction and k enrollment utterance-label pairs, then processed by an instruction-tuned encoder-decoder LLM that generates the target emotion token. MetaICL is a two-stage fine-tuning: first the speech LM is adapted to SER without enrollment examples, then it is trained per speaker to predict target emotions conditioned on randomly sampled enrollment pairs under target-uncontrolled and label-uncontrolled settings. This teaches the model to do in-context learning for speaker adaptatio

What would settle it

Run the proposed 0-shot and 7-shot ICL evaluation on a public SER benchmark such as IEMOCAP using a model meta-trained on the paper's protocol, and check whether in-context personalization still outperforms the 0-shot baseline and enrollment baselines by a comparable margin. Alternatively, an annotation study testing whether human raters reliably distinguish the seven recorded portrayals within each speaker would directly test the dataset's grounding.

Watch

Extended reading notes

Core claim

The central discovery is that an LLM extended to accept speech can be meta-trained to perform in-context learning for speaker personalization rather than task adaptation. By randomly sampling zero to seven enrollment utterance-label pairs from the same speaker's own emotion set (target-uncontrolled, label-uncontrolled) and fine-tuning with the MetaICL objective, the model learns to use arbitrary enrollment examples as conditioning. At test time, the authors feed k-shot enrollment pairs of an unseen target speaker; each added shot raises per-speaker unweighted accuracy, with the best configuration (7-shot, target-overlap and label-disjoint) reaching 0.757 versus 0.675 for 0-shot, and outperfo

Load-bearing premise

The claim rests on the assumption that the newly collected acted emotional recordings, in which each of 800 speakers was instructed to speak 'as clearly as possible so that your emotions come across', represent genuine, consistent, speaker-specific emotional expression; no annotator agreement, cross-dataset validation, or comparison on an existing public SER benchmark is provided to verify that assumption.

Editorial extensions

If this is right

  • Adding seven enrollment utterance-label pairs to a MetaICL-trained speech LM raises per-speaker unweighted accuracy from 0.675 to 0.757, an 8.2-point gain, on an unseen-speaker evaluation.
  • The improvement holds under multiple enrollment-label configurations (target-overlap vs exclusive, label-disjoint vs unrestricted); only redundant-label settings (TU+LO) saturate at two shots.
  • Meta-training with varied emotion labels (TU+LU 0:7) gives better few-shot personalization than meta-training with a fixed 7-shot full-emotion set or neutral-only examples, as shown in Table IV.
  • The proposed speech LM also beats the best classifier personalization baseline (Personalized A, 0.626), indicating that the generative speech-LM formulation is the better carrier of speaker-specific cues.

Reading between the lines

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

  • Inference: Because the paper's data is acted Japanese speech, the same protocol should be tested on spontaneous public corpora before the gains are claimed to generalize.
  • Inference: The monotonic improvement up to seven shots is not shown to saturate; scaling shots further or adding enrollment sessions could extend the result.
  • Inference: The in-context conditioning could be combined with textual speaker descriptors (age, gender, personality) in the same prompt, since the prompt already mixes speech and text tokens.
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 / 6 minor

Summary. The paper proposes to personalize speech emotion recognition (SER) for unseen speakers by conditioning a speech-language model on a few enrollment utterance-label pairs of the target speaker via in-context learning (ICL). The model is meta-trained with MetaICL on a newly collected 800-speaker Japanese acted-emotion dataset. Experiments compare classifier baselines and speech-LM zero-shot inference, reporting that the proposed 7-shot TO+LD condition improves UA_spk from 0.675 (0-shot speech LM) to 0.757, with ablations on enrollment selection and meta-training input.

Significance. If the result holds, the paper addresses a real and timely problem: speaker adaptation for SER with flexible enrollment, using the ICL capability of speech LMs. The strengths are the held-out-speaker evaluation, the explicit exclusion of the target utterance from the enrollment set, the inclusion of multiple ICL settings (TU+LD, TE+LU, TU+LO, etc.), and ablations on the meta-training input in Table IV. These design choices make the reported gains genuine held-out predictions rather than circular fitting. However, the significance is currently limited by the fact that all empirical support comes from a single newly collected, non-public, acted dataset with no human annotation validation, no public benchmark comparison, and no statistical significance testing. The central claim therefore needs stronger external validation before it can be generalized.

major comments (4)
  1. [Section V-A and V-D] The sole empirical support for the claim is a newly collected, non-public, acted dataset. No human annotation agreement is reported, and no cross-corpus evaluation is performed. The recording protocol (same headset, laptop, quiet conference room, one session per speaker) means enrollment and target utterances may share stable session/channel characteristics, so the ICL gain could reflect channel identification rather than speaker-specific emotion expression. Please add either (i) human annotation agreement for the acted portrayals, (ii) cross-dataset evaluation on a public SER corpus (e.g., IEMOCAP, CREMA-D) with an analogous enrollment protocol, or (iii) an analysis showing that channel/session cues are not responsible for the improvement. This is load-bearing for the advertised real-world applicability.
  2. [Tables II and III, Section V-E] No confidence intervals, paired significance tests, or effect sizes are reported for any of the accuracy comparisons. For instance, several adjacent k-shot differences are about 0.01–0.02 while the reported per-speaker standard deviation is around 0.10, so the claim that 'performance improved as the number of enrollment utterance-label pairs increased' is not established for every step. The main 0.757 vs 0.675 comparison may be significant with the large test set, but the paper should report per-speaker paired statistics (e.g., Wilcoxon or bootstrap CIs) to support the incremental gains and the comparisons in Tables III and IV.
  3. [Table II and Section V-E] The headline result is the 7-shot TO+LD condition, which uses one enrollment utterance for each of the seven emotion labels. This is exactly the fixed all-emotion enrollment scenario that the introduction argues is difficult to prepare. The flexibility advantage of the proposed method therefore rests on the smaller gains in the TU+LD/TU+LU conditions, not on the strongest reported number. The paper should either report the best label-flexible result as the primary claim or explicitly discuss the trade-off between flexibility and accuracy.
  4. [Section V-B, Table II] The comparison to 'conventional methods' such as Personalized A is confounded by model capacity: the proposed method uses a 0.6B-parameter LLM plus trainable Q-Former, while the classifier baselines use a much smaller speech encoder. The more informative same-architecture comparison is the proposed method vs the speech LM 0-shot, which does show a clear improvement. The claim that the proposed method 'outperforms conventional methods' should be qualified to acknowledge this capacity difference, or controlled by scaling the classifier baseline.
minor comments (6)
  1. [Section IV-A] The text says the first fine-tuning stage is 'as in Eq. (15)', but Eq. (15) defines the decoder transform F; the intended reference is likely Eq. (16), the training objective. Please correct.
  2. [Equation (7)] The objective for instruction-tuning is written as minimizing P(O|W; Θ_llm); it should be minimizing the negative log-likelihood, as in Eq. (16). This is a typo but affects readability.
  3. [Section V-E] The name '0-shot' is used twice in Table II with different meanings: the non-MetaICL speech LM trained with Eq. (16) and the proposed MetaICL model at zero ICL examples. Please disambiguate, e.g., 'SLM 0-shot' vs 'Proposed 0-shot'.
  4. [Section V-C] The Q-Former query length (150) is said to be set 'based on preliminary experiments', but no sensitivity analysis is provided. A brief ablation or a reference to a supplementary would help.
  5. [Section IV-A] The notation '0:7' in 'TU+LU 0:7' is used without an explicit definition; please spell out that k is drawn from 0 to 7.
  6. [Section IV-A] There are typos in the definitions of LO and LD: 'same emotion emotion' and 'different emotion emotions' should be 'same emotion' and 'different emotions'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's central claim rests on held-out speaker evaluation of a meta-trained speech LM, not on a fitted quantity or self-citation chain.

full rationale

The paper proposes a MetaICL-trained speech LM for personalized speech emotion recognition via in-context learning. The core result is an empirical comparison on held-out speakers: train/test speakers are disjoint by construction, and the enrollment utterance-label pairs used at inference come from target speakers not seen during meta-training. The training objective in Eq. (18) is a standard MetaICL loss over training speakers, and inference in Eq. (20) conditions on target-speaker examples; neither equation defines the reported accuracy in terms of a fitted parameter or a self-referential identity. The improvements in UA_spk from 0.675 (0-shot) to 0.757 (7-shot) are genuine held-out predictions, not a renamed training objective. There is no reliance on a uniqueness theorem or on a load-bearing self-citation: the cited prior work is used for standard components (MetaICL, Q-Former, speech LM architectures) and does not itself contain the paper's claim. The main concern is that the newly collected acted dataset has no external validation or cross-dataset comparison, and same-session enrollment could in principle allow channel/session artifacts to be exploited. However, that is a question of dataset validity and external generalization, not circularity in the derivation chain. No circular step was identified.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities or forces. It relies on standard pretrained model assumptions and a self-collected acted dataset. The single hand-tuned hyperparameter is the Q-Former query length; all other training hyperparameters (dropout, label smoothing, beam size) are standard and not central to the claim.

free parameters (1)
  • Q-Former query length = 150
    Section V-C: 'Based on preliminary experiments, the length of the query was set to 150, as it yielded the best performance.' This is a hand-tuned architectural hyperparameter that affects how much speech information is passed to the LLM.
assumptions (4)
  • domain assumption Pretrained speech encoder and 0.6B LLM transfer useful representations to SER after fine-tuning
    Section V-B: the speech encoder is pre-trained on 'various speech-understanding tasks' and the LLM is pre-trained on a large text corpus and instruction-tuned. The paper assumes these representations are suitable for emotion recognition after meta-training, without comparing to other backbones.
  • domain assumption The seven acted emotion categories (anger, disgust, fear, joy, sadness, surprise, neutral) are distinct and consistently performed by all speakers
    Section V-A: speakers were instructed to 'speak as clearly as possible so that your emotions come across'. The validity of these portrayals is not verified by annotator agreement or acoustic analysis.
  • domain assumption Enrollment utterances from a target speaker are representative of that speaker's emotional expression and can be used to personalize inference
    Section IV-A: the method conditions on enrollment utterance-label pairs from the same speaker as the target speech. This assumes speaker-specific emotional expression is consistent enough across different utterances to transfer.
  • domain assumption The LLM's in-context learning capability, developed for text, transfers to speech inputs through the Q-Former adapter
    Section III-B and IV-A: the Q-Former converts speech representations into queries that are concatenated with text embeddings. The paper relies on the LLM being able to interpret these cross-modal sequences as ICL demonstrations, which is not proven a priori.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Few-shot Personalization via In-Context Learning for Speech Emotion Recognition based on Speech-Language Model." pith.science (2026). https://pith.science/paper/CGGJBI7B

@misc{pith2026250908344,
  author       = {Pith},
  title        = {Pith review of: Few-shot Personalization via In-Context Learning for Speech Emotion Recognition based on Speech-Language Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CGGJBI7B}},
  note         = {Machine review of arXiv:2509.08344}
}
read the original abstract

This paper proposes a personalization method for speech emotion recognition (SER) through in-context learning (ICL). Since the expression of emotions varies from person to person, speaker-specific adaptation is crucial for improving the SER performance. Conventional SER methods have been personalized using emotional utterances of a target speaker, but it is often difficult to prepare utterances corresponding to all emotion labels in advance. Our idea to overcome this difficulty is to obtain speaker characteristics by conditioning a few emotional utterances of the target speaker in ICL-based inference. ICL is a method to perform unseen tasks by conditioning a few input-output examples through inference in large language models (LLMs). We meta-train a speech-language model extended from the LLM to learn how to perform personalized SER via ICL. Experimental results using our newly collected SER dataset demonstrate that the proposed method outperforms conventional methods.

Figures

Figures reproduced from arXiv: 2509.08344 by the authors.

Figure 1
Figure 1. Example of personalization for SER via ICL in speech LM. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of classifier model, speech language model, and proposed method for SER. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. LLM-Based Synthetic Ground Truth Generation for Audio-Based Emotion Classification via In-Context Learning

    cs.SD 2026-06 unverdicted novelty 3.0 of 10

    Proposes an LLM agentic workflow with retrieval-based in-context learning to generate synthetic ground truth for audio emotion classification in multi-user VR environments.

Reference graph

Works this paper leans on

45 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    Customer Satisfaction Estimation in Contact Center Calls Based on a Hierarchical Multi-Task Model,

    A. Ando, R. Masumura, H. Kamiyama, S. Kobashikawa, Y . Aono, and T. Toda, “Customer Satisfaction Estimation in Contact Center Calls Based on a Hierarchical Multi-Task Model,”IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLPRO), pp. 715–728, 2020

    work page 2020

  2. [2]

    Using Emotion to Gain Rapport in a Spoken Dialog System,

    J. C. Acosta, “Using Emotion to Gain Rapport in a Spoken Dialog System,” inProc. North American Chapter of the Association for Computational Linguistics (NAACL), 2009, p. 49–54

    work page 2009

  3. [3]

    Affect intensity as an individual difference characteristic: A review,

    R. J. Larsen and E. Diener, “Affect intensity as an individual difference characteristic: A review,”Journal of Research in personality, pp. 1–39, 1987

    work page 1987

  4. [4]

    Is There Universal Recognition of Emotion from Facial Expression? A Review of the Cross-Cultural Studies,

    J. A. Russell, “Is There Universal Recognition of Emotion from Facial Expression? A Review of the Cross-Cultural Studies,”Psychological bulletin, p. 102, 1994

    work page 1994

  5. [5]

    Emotion and Aging: Experience, Expression, and Control

    J. J. Gross, L. L. Carstensen, M. Pasupathi, J. Tsai, C. G ¨otestam Skorpen, and A. Y . Hsu, “Emotion and Aging: Experience, Expression, and Control.”Psychology and aging, p. 590, 1997

    work page 1997

  6. [6]

    Gender Differences in Emotion Expression in Children: a Meta-Analytic Review,

    T. M. Chaplin and A. Aldao, “Gender Differences in Emotion Expression in Children: a Meta-Analytic Review,”Psychological bulletin, p. 735, 2013

    work page 2013

  7. [7]

    The Independent Effects of Personality and Situations on Real- time Expressions of Behavior and Emotion,

    R. A. Sherman, J. F. Rauthmann, N. A. Brown, D. G. Serfass, and A. B. Jones, “The Independent Effects of Personality and Situations on Real- time Expressions of Behavior and Emotion,”Journal of personality and social psychology, p. 872, 2015

    work page 2015

  8. [8]

    Domain Invariant Feature Learning for Speaker-Independent Speech Emotion Recognition,

    C. Lu, Y . Zong, W. Zheng, Y . Li, C. Tang, and B. W. Schuller, “Domain Invariant Feature Learning for Speaker-Independent Speech Emotion Recognition,”IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLPRO), pp. 2217–2230, 2022

    work page 2022

Show all 45 references
  1. [9]

    SHEEP, GOATS, LAMBS and WOLVES: a statistical analysis of speaker performance in the NIST 1998 speaker recognition evaluation,

    G. R. Doddington, W. Liggett, A. F. Martin, M. A. Przybocki, and D. A. Reynolds, “SHEEP, GOATS, LAMBS and WOLVES: a statistical analysis of speaker performance in the NIST 1998 speaker recognition evaluation,” inProc. International Conference on Spoken Language Processing (ICS...

    1998

  2. [10]

    ISNet: Individual Standardization Network for Speech Emotion Recognition,

    W. Fan, X. Xu, B. Cai, and X. Xing, “ISNet: Individual Standardization Network for Speech Emotion Recognition,”IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLPRO), pp. 1803–1814, 2022

    2022

  3. [11]

    Deep speaker con- ditioning for speech emotion recognition,

    A. Triantafyllopoulos, S. Liu, and B. W. Schuller, “Deep speaker con- ditioning for speech emotion recognition,” inProc. IEEE International Conference on Multimedia and Expo (ICME), 2021, pp. 1–6

    2021

  4. [12]

    Exploring speaker enrolment for few-shot personalisation in emotional vocalisation prediction,

    A. Triantafyllopoulos, M. Song, Z. Yang, X. Jing, and B. W. Schuller, “Exploring speaker enrolment for few-shot personalisation in emotional vocalisation prediction,”arXiv preprint arXiv:2206.06680, 2022

    2022 arXiv

  5. [13]

    Enrolment-based personalisation for improving individual-level fairness in speech emotion recognition,

    A. Triantafyllopoulos and B. Schuller, “Enrolment-based personalisation for improving individual-level fairness in speech emotion recognition,” inProc. Conference of the International Speech Communication Asso- ciation (INTERSPEECH), 2024, pp. 3729–3733

    2024

  6. [14]

    Language Models are Few-Shot Learners,

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert- V oss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. C...

    2020

  7. [15]

    Steering Language Model to Stable Speech Emotion Recognition via Contextual Perception and Chain of Thought,

    Z. Zhao, X. Zhu, X. Wang, S. Wang, X. Geng, W. Tian, and L. Xie, “Steering Language Model to Stable Speech Emotion Recognition via Contextual Perception and Chain of Thought,”arXiv preprint arXiv:2502.18186, 2025

    2025

  8. [16]

    SECap: Speech Emotion Captioning with Large Language Model,

    Y . Xu, H. Chen, J. Yu, Q. Huang, Z. Wu, S.-X. Zhang, G. Li, Y . Luo, and R. Gu, “SECap: Speech Emotion Captioning with Large Language Model,” inProc. Association for the Advancement of Artificial Intelligence (AAAI), 2024, pp. 19 323–19 331

    2024

  9. [17]

    MetaICL: Learning to learn in context,

    S. Min, M. Lewis, L. Zettlemoyer, and H. Hajishirzi, “MetaICL: Learning to learn in context,” inProc. North American Chapter of the Association for Computational Linguistics (NAACL), 2022, pp. 2791– 2809

    2022

  10. [18]

    Dawn of the Transformer Era in Speech Emotion Recognition: Closing the Valence Gap,

    J. Wagner, A. Triantafyllopoulos, H. Wierstorf, M. Schmitt, F. Burkhardt, F. Eyben, and B. W. Schuller, “Dawn of the Transformer Era in Speech Emotion Recognition: Closing the Valence Gap,”IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 10 745–10 759, 2023

    2023

  11. [19]

    Multimodal Speech Emotion Recogni- tion Using Audio and Text,

    S. Yoon, S. Byun, and K. Jung, “Multimodal Speech Emotion Recogni- tion Using Audio and Text,” inProc. IEEE Spoken Language Technology Workshop (SLT), 2018, pp. 112–118

    2018

  12. [20]

    Speech Emotion Recognition with Dual-Sequence LSTM Architecture,

    J. Wang, M. Xue, R. Culhane, E. Diao, J. Ding, and V . Tarokh, “Speech Emotion Recognition with Dual-Sequence LSTM Architecture,” inProc. IEEE International Conference on Acoustics, Speech and Signal Pro- cessing (ICASSP), 2020, pp. 6474–6478

    2020

  13. [21]

    Removing Bias with Residual Mixture of Multi-view Attention for Speech Emotion Recog- nition,

    M. A. Jalal, R. Milner, T. Hain, and R. K. Moore, “Removing Bias with Residual Mixture of Multi-view Attention for Speech Emotion Recog- nition,” inProc. Conference of the International Speech Communication Association (INTERSPEECH), 2020, pp. 4084–4088

    2020

  14. [22]

    Speaker Independent Emotion Recognition by Early Fusion of Acoustic and Linguistic Features Within Ensembles,

    B. Schuller, R. M ¨uller, M. Lang, and G. Rigoll, “Speaker Independent Emotion Recognition by Early Fusion of Acoustic and Linguistic Features Within Ensembles,” 2005

    2005

  15. [23]

    Speaker Independent Speech Emotion Recognition by Ensemble Clas- sification,

    B. Schuller, S. Reiter, R. Muller, M. Al-Hames, M. Lang, and G. Rigoll, “Speaker Independent Speech Emotion Recognition by Ensemble Clas- sification,” inProc. IEEE International Conference on Multimedia and Expo (ICME), 2005, pp. 864–867

    2005

  16. [24]

    Emotions are A Personal Thing: Towards Speaker-Adaptive Emotion Recognition,

    M. Sidorov, S. Ultes, and A. Schmitt, “Emotions are A Personal Thing: Towards Speaker-Adaptive Emotion Recognition,” inProc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2014, pp. 4803–4807

    2014

  17. [25]

    Learning Utterance-Level Representations for Speech Emotion and Age/Gender Recognition Using Deep Neural Networks,

    Z.-Q. Wang and I. Tashev, “Learning Utterance-Level Representations for Speech Emotion and Age/Gender Recognition Using Deep Neural Networks,” inProc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2017, pp. 5150–5154

    2017

  18. [26]

    Gender-aware CNN-BLSTM for speech emotion recognition,

    L. Zhang, L. Wang, J. Dang, L. Guo, and Q. Yu, “Gender-aware CNN-BLSTM for speech emotion recognition,” inProc. International Conference on Artificial Neural Networks (ICANN), 2018, pp. 782–790

    2018

  19. [27]

    Improved End-to-End Speech Emotion Recognition using Self Attention Mechanism and Multitask Learning,

    Y . Li, T. Zhao, T. Kawaharaet al., “Improved End-to-End Speech Emotion Recognition using Self Attention Mechanism and Multitask Learning,” inProc. Conference of the International Speech Communi- cation Association (INTERSPEECH), 2019, pp. 2803–2807

    2019

  20. [28]

    Speaker Attentive Speech Emotion Recognition,

    C. Le Moine, N. Obin, and A. Roebel, “Speaker Attentive Speech Emotion Recognition,” inProc. Conference of the International Speech Communication Association (INTERSPEECH), 2021, pp. 2866–2870

    2021

  21. [29]

    Individual-Aware Attention Modulation for Unseen Speaker Emotion Recognition,

    Y . Fang, X. Xing, Z. Chu, Y . Du, and X. Xu, “Individual-Aware Attention Modulation for Unseen Speaker Emotion Recognition,”IEEE Transactions on Affective Computing, pp. 1–14, 2024

    2024

  22. [30]

    Speaker Normalization for Self-Supervised Speech Emotion Recognition,

    I. Gat, H. Aronowitz, W. Zhu, E. Morais, and R. Hoory, “Speaker Normalization for Self-Supervised Speech Emotion Recognition,” in Proc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022, pp. 7342–7346

    2022

  23. [31]

    SALMONN: Towards Generic Hearing Abilities for Large Language Models,

    T. Changli, Y . Wenyi, S. Guangzhi, C. Xianzhao, T. Tian, L. Wei, L. Lu, M. Zejun, and Z. Chao, “SALMONN: Towards Generic Hearing Abilities for Large Language Models,”arXiv preprint arXiv:2310.13289, 2023

    2023 arXiv

  24. [32]

    Audio Flamingo: A Novel Audio Language Model with Few-Shot Learning and Dialogue Abilities,

    Z. Kong, A. Goel, R. Badlani, W. Ping, R. Valle, and B. Catanzaro, “Audio Flamingo: A Novel Audio Language Model with Few-Shot Learning and Dialogue Abilities,” inProc. International Conference on Machine Learning (ICML), 2024, pp. 25 125–25 148

    2024

  25. [33]

    Qwen2-Audio Technical Report,

    Y . Chu, J. Xu, Q. Yang, H. Wei, X. Wei, Z. Guo, Y . Leng, Y . Lv, J. He, J. Lin, C. Zhou, and J. Zhou, “Qwen2-Audio Technical Report,”arXiv preprint arXiv:2407.10759, 2024

    2024 arXiv

  26. [34]

    SALM: Speech-Augmented Language Model with In-Context Learning for Speech Recognition and Translation,

    Z. Chen, H. Huang, A. Andrusenko, O. Hrinchuk, K. C. Puvvada, J. Li, S. Ghosh, J. Balam, and B. Ginsburg, “SALM: Speech-Augmented Language Model with In-Context Learning for Speech Recognition and Translation,” inProc. IEEE International Conference on Acoustics, Speech and Sig...

    2024

  27. [35]

    Can Whisper Perform Speech-Based In-Context Learning?

    S. Wang, C.-H. Yang, J. Wu, and C. Zhang, “Can Whisper Perform Speech-Based In-Context Learning?” inProc. IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP), 2024, pp. 13 421–13 425

    2024

  28. [36]

    In-Context Learning Boosts Speech Recognition via Human- like Adaptation to Speakers and Language Varieties,

    N. Roll, C. Graham, Y . Tatsumi, K. T. Nguyen, M. Sumner, and D. Ju- rafsky, “In-Context Learning Boosts Speech Recognition via Human- like Adaptation to Speakers and Language Varieties,”arXiv preprint arXiv:2505.14887, 2025

    2025 arXiv

  29. [37]

    COSMIC: Data Efficient Instruction-tuning For Speech In-Context Learning,

    J. Pan, J. Wu, Y . Gaur, S. Sivasankaran, Z. Chen, S. Liu, and J. Li, “COSMIC: Data Efficient Instruction-tuning For Speech In-Context Learning,” inProc. Conference of the International Speech Commu- nication Association (INTERSPEECH), 2024, pp. 4164–4168

    2024

  30. [38]

    Exploring In- Context Learning of Textless Speech Language Model for Speech Classification Tasks,

    K.-W. Chang, H. Ming-Hao, S.-W. Li, and H.-y. Lee, “Exploring In- Context Learning of Textless Speech Language Model for Speech Classification Tasks,” inProc. Conference of the International Speech Communication Association (INTERSPEECH), 2024, pp. 4139–4143

    2024

  31. [39]

    Attention is All You Need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is All You Need,” inProc. Advances in neural information processing systems (NIPS), 2017, pp. 5998–6008

    2017

  32. [40]

    Finetuned Language Models are Zero-Shot Learners,

    J. Wei, M. Bosma, V . Zhao, K. Guu, A. W. Yu, B. Lester, N. Du, A. M. Dai, and Q. V . Le, “Finetuned Language Models are Zero-Shot Learners,” inProc. International Conference on Learning Representa- tions (ICLR), 2022

    2022

  33. [41]

    BLIP-2: Bootstrapping Language- Image Pre-training with Frozen Image Encoders and Large Language Models,

    J. Li, D. Li, S. Savarese, and S. Hoi, “BLIP-2: Bootstrapping Language- Image Pre-training with Frozen Image Encoders and Large Language Models,” inProc. International Conference on Machine Learning (ICML), 2023, pp. 19 730–19 742

    2023

  34. [42]

    IEMOCAP: Interactive emotional dyadic motion capture database,

    C. Busso, M. Bulut, C.-C. Lee, A. Kazemzadeh, E. Mower, S. Kim, J. N. Chang, S. Lee, and S. S. Narayanan, “IEMOCAP: Interactive emotional dyadic motion capture database,” inProc. Language Resources and Evaluation Conference (LREC), 2008, pp. 335–359

    2008

  35. [43]

    A database of German emotional speech,

    F. Burkhardt, A. Paeschke, M. Rolfes, W. F. Sendlmeier, B. Weiss et al., “A database of German emotional speech,” inProc. Conference of the International Speech Communication Association (INTERSPEECH), 2005, pp. 1517–1520

    2005

  36. [44]

    Speaker-Dependent Audio-Visual Emotion Recognition,

    S. Haq, P. J. Jackson, and J. Edge, “Speaker-Dependent Audio-Visual Emotion Recognition,” inProc. International Conference on Auditory- Visual Speech Processing (AVSP), vol. 2009, 2009, pp. 53–58

    2009

  37. [45]

    Ul2: Unifying language learning paradigms,

    Y . Tay, M. Dehghani, V . Q. Tran, X. Garcia, J. Wei, X. Wang, H. W. Chung, D. Bahri, T. Schuster, S. Zhenget al., “Ul2: Unifying language learning paradigms,” inProc. International Conference on Learning Representations (ICLR), 2023

    2023

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.