{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,17]],"date-time":"2026-06-17T13:57:54Z","timestamp":1781704674477,"version":"3.54.5"},"reference-count":117,"publisher":"MIT Press","license":[{"start":{"date-parts":[[2026,6,17]],"date-time":"2026-06-17T00:00:00Z","timestamp":1781654400000},"content-version":"vor","delay-in-days":167,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2026,6,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Spurred by the demand for interpretable models, research on explainable AI for language technologies has experienced significant growth, with feature attribution methods emerging as a cornerstone of this progress. While prior work in NLP explored such methods for classification tasks and textual applications, explainability intersecting generation and speech is lagging, with existing techniques failing to account for the autoregressive nature of state-of-the-art models and to provide finegrained, phonetically meaningful explanations. We address this gap by introducing Spectrogram Perturbation for Explainable Speech-to-text Generation (SPES), a feature attribution technique applicable to sequence generation tasks with autoregressive models. SPES provides explanations for each predicted token based on both the input spectrogram and the previously generated tokens. Extensive evaluation on speech recognition and translation demonstrates that SPES generates explanations that are faithful and plausible to humans.<\/jats:p>","DOI":"10.1162\/tacl.a.684","type":"journal-article","created":{"date-parts":[[2026,6,17]],"date-time":"2026-06-17T13:27:18Z","timestamp":1781702838000},"page":"772-801","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":0,"title":["SPES: Spectrogram Perturbation for Explainable Speech-to-Text Generation"],"prefix":"10.1162","volume":"14","author":[{"given":"Dennis","family":"Fucci","sequence":"first","affiliation":[{"name":"Fondazione Bruno Kessler, Trento, Italy, dennisfucci@gmail.com"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marco","family":"Gaido","sequence":"additional","affiliation":[{"name":"Fondazione Bruno Kessler, Trento, Italy, mgaido@fbk.eu"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Beatrice","family":"Savoldi","sequence":"additional","affiliation":[{"name":"Fondazione Bruno Kessler, Trento, Italy, bsavoldi@fbk.eu"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Matteo","family":"Negri","sequence":"additional","affiliation":[{"name":"Fondazione Bruno Kessler, Trento, Italy, negri@fbk.eu"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mauro","family":"Cettolo","sequence":"additional","affiliation":[{"name":"Fondazione Bruno Kessler, Trento, Italy, cettolo@fbk.eu"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Luisa","family":"Bentivogli","sequence":"additional","affiliation":[{"name":"Fondazione Bruno Kessler, Trento, Italy, bentivo@fbk.eu"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"281","published-online":{"date-parts":[[2026,6,1]]},"reference":[{"issue":"11","key":"2026061709271310100_bib1","doi-asserted-by":"crossref","first-page":"2274","DOI":"10.1109\/TPAMI.2012.120","article-title":"SLIC Superpixels Compared to State-of-the-Art Superpixel Methods","volume":"34","author":"Achanta","year":"2012","journal-title":"IEEE Transactions on Pattern Analysis and Machine Intelligence"},{"key":"2026061709271310100_bib2","article-title":"Towards better understanding of gradient-based attribution methods for Deep Neural Networks","volume-title":"Proceedings of the 6th International Conference on Learning Representations","author":"Ancona","year":"2018"},{"key":"2026061709271310100_bib3","first-page":"4218","article-title":"Common Voice: A Massively-Multilingual Speech Corpus","volume-title":"Proceedings of the Twelfth Language Resources and Evaluation Conference","author":"Ardila","year":"2020"},{"key":"2026061709271310100_bib4","first-page":"159","article-title":"Explaining Recurrent Neural Network Predictions in Sentiment Analysis","volume-title":"Proceedings of the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis","author":"Arras","year":"2017"},{"key":"2026061709271310100_bib5","doi-asserted-by":"crossref","first-page":"21318","DOI":"10.18653\/v1\/2024.emnlp-main.1188","article-title":"Twists, humps, and pebbles: Multilingual speech recognition models exhibit gender performance gaps","volume-title":"Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing","author":"Attanasio","year":"2024"},{"key":"2026061709271310100_bib6","doi-asserted-by":"crossref","first-page":"23181","DOI":"10.18653\/v1\/2025.findings-acl.1190","article-title":"Lost in transcription, found in distribution shift: Demystifying hallucination in speech foundation models","volume-title":"Findings of the Association for Computational Linguistics: ACL 2025","author":"Atwany","year":"2025"},{"issue":"7","key":"2026061709271310100_bib7","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1371\/journal.pone.0130140","article-title":"On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation","volume":"10","author":"Bach","year":"2015","journal-title":"PLOS ONE"},{"key":"2026061709271310100_bib8","doi-asserted-by":"crossref","first-page":"82","DOI":"10.1016\/j.inffus.2019.12.012","article-title":"Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI","volume":"58","author":"Arrieta","year":"2020","journal-title":"Information Fusion"},{"issue":"1","key":"2026061709271310100_bib9","doi-asserted-by":"crossref","first-page":"418","DOI":"10.1016\/j.jfranklin.2023.11.038","article-title":"AudioMNIST: Exploring Explainable Artificial Intelligence for audio analysis on a simple benchmark","volume":"361","author":"Becker","year":"2024","journal-title":"Journal of the Franklin Institute"},{"key":"2026061709271310100_bib10","doi-asserted-by":"crossref","first-page":"6923","DOI":"10.18653\/v1\/2020.acl-main.619","article-title":"Gender in Danger? Evaluating Speech Translation Technology on the MuST-SHE Corpus","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Bentivogli","year":"2020"},{"key":"2026061709271310100_bib11","volume-title":"On the Opportunities and Risks of Foundation Models","author":"Bommasani","year":"2021"},{"key":"2026061709271310100_bib12","doi-asserted-by":"crossref","first-page":"131","DOI":"10.1016\/j.patrec.2023.10.012","article-title":"Feature perturbation augmentation for reliable evaluation of importance estimators in neural networks","volume":"176","author":"Brocki","year":"2023","journal-title":"Pattern Recognition Letters"},{"issue":"8","key":"2026061709271310100_bib13","doi-asserted-by":"crossref","DOI":"10.3390\/electronics8080832","article-title":"Machine Learning Interpretability: A Survey on Methods and Metrics","volume":"8","author":"Carvalho","year":"2019","journal-title":"Electronics"},{"key":"2026061709271310100_bib14","doi-asserted-by":"crossref","first-page":"180","DOI":"10.1016\/j.cor.2017.01.019","article-title":"Improving polynomial estimation of the shapley value by stratified random sampling with optimum allocation","volume":"82","author":"Castro","year":"2017","journal-title":"Computers & Operations Research"},{"key":"2026061709271310100_bib15","doi-asserted-by":"crossref","first-page":"101155","DOI":"10.1016\/j.csl.2020.101155","article-title":"MuST-C: A multilingual corpus for end-to-end speech translation","volume":"66","author":"Cattoni","year":"2021","journal-title":"Computer Speech & Language"},{"key":"2026061709271310100_bib16","doi-asserted-by":"crossref","first-page":"102135","DOI":"10.1016\/j.tele.2024.102135","article-title":"EXplainable Artificial Intelligence (XAI) for facilitating recognition of algorithmic bias: An experiment from imposed users\u2019 perspectives","volume":"91","author":"Chuan","year":"2024","journal-title":"Telematics and Informatics"},{"key":"2026061709271310100_bib17","volume-title":"Efficient XAI Techniques: A Taxonomic Survey","author":"Chuang","year":"2023"},{"key":"2026061709271310100_bib18","volume-title":"CoRTX: Contrastive Framework for Real-time Explanation","author":"Chuang","year":"2023"},{"key":"2026061709271310100_bib19","volume-title":"SeamlessM4T: Massively Multilingual & Multimodal Machine Translation","author":"Communication","year":"2023"},{"key":"2026061709271310100_bib20","doi-asserted-by":"crossref","first-page":"16318","DOI":"10.52202\/075280-0719","article-title":"Towards Automated Circuit Discovery for Mechanistic Interpretability","volume":"36","author":"Conmy","year":"2023","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2026061709271310100_bib21","article-title":"Understanding global feature contributions with additive importance measures","volume":"33","author":"Covert","year":"2020","journal-title":"Advances on Neural Information Processing Systems"},{"issue":"1","key":"2026061709271310100_bib22","first-page":"9477","article-title":"Explaining by removing: a unified framework for model explanation","volume":"22","author":"Covert","year":"2021","journal-title":"The Journal of Machine Learning Research"},{"key":"2026061709271310100_bib23","article-title":"Real Time Image Saliency for Black Box Classifiers","volume-title":"Advances in Neural Information Processing Systems","author":"Dabkowski","year":"2017"},{"key":"2026061709271310100_bib24","doi-asserted-by":"crossref","first-page":"447","DOI":"10.18653\/v1\/2020.aacl-main.46","article-title":"A Survey of the State of Explainable AI for Natural Language Processing","volume-title":"Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing","author":"Danilevsky","year":"2020"},{"key":"2026061709271310100_bib25","first-page":"137","article-title":"On Target Segmentation for Direct Speech Translation","volume-title":"Proceedings of the 14th Conference of the Association for Machine Translation in the Americas (Volume 1: Research Track)","author":"Di Gangi","year":"2020"},{"key":"2026061709271310100_bib26","doi-asserted-by":"crossref","DOI":"10.1016\/j.patcog.2021.108194","article-title":"Towards robust explanations for deep neural networks","volume":"121","author":"Dombrowski","year":"2022","journal-title":"Pattern Recognition"},{"key":"2026061709271310100_bib27","volume-title":"Towards A Rigorous Science of Interpretable Machine Learning","author":"Doshi-Velez","year":"2017"},{"key":"2026061709271310100_bib28","volume-title":"Risks and Opportunities of Open-Source Generative AI","author":"Eiras","year":"2024"},{"key":"2026061709271310100_bib29","doi-asserted-by":"crossref","first-page":"16153","DOI":"10.1109\/CVPR52729.2023.01550","article-title":"Don\u2019t Lie to Me! Robust and Efficient Explainability with Verified Perturbation Analysis","volume-title":"2023 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Fel","year":"2023"},{"issue":"1","key":"2026061709271310100_bib30","doi-asserted-by":"crossref","first-page":"2095039","DOI":"10.1080\/08839514.2022.2095039","article-title":"Automatic Speech Recognition Using Limited Vocabulary: A Survey","volume":"36","author":"Fendji","year":"2022","journal-title":"Applied Artificial Intelligence"},{"key":"2026061709271310100_bib31","volume-title":"A Primer on the Inner Workings of Transformer-based Language Models","author":"Ferrando","year":"2024"},{"key":"2026061709271310100_bib32","doi-asserted-by":"crossref","first-page":"17432","DOI":"10.18653\/v1\/2024.emnlp-main.965","article-title":"Information flow routes: Automatically interpreting language models at scale","volume-title":"Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing","author":"Ferrando","year":"2024"},{"key":"2026061709271310100_bib33","first-page":"3449","article-title":"Interpretable Explanations of Black Boxes by Meaningful Perturbation","volume-title":"Proceedings of the IEEE International Conference on Computer Vision (ICCV)","author":"Fong","year":"2017"},{"key":"2026061709271310100_bib34","volume-title":"Hallucinations in neural automatic speech recognition: Identifying errors and hallucinatory models","author":"Frieske","year":"2024"},{"key":"2026061709271310100_bib35","first-page":"373","article-title":"Explainability for Speech Models: On the Challenges of Acoustic Feature Selection","volume-title":"Proceedings of the 10th Italian Conference on Computational Linguistics (CLiC-it 2024)","author":"Fucci","year":"2024"},{"issue":"9","key":"2026061709271310100_bib36","doi-asserted-by":"crossref","first-page":"6019","DOI":"10.1007\/s10994-024-06550-x","article-title":"Evaluating feature attribution methods in the image domain","volume":"113","author":"Gevaert","year":"2024","journal-title":"Machine Learning"},{"key":"2026061709271310100_bib37","doi-asserted-by":"crossref","first-page":"369","DOI":"10.1145\/1143844.1143891","article-title":"Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks","volume-title":"Proceedings of the 23rd International Conference on Machine Learning","author":"Graves","year":"2006"},{"key":"2026061709271310100_bib38","doi-asserted-by":"crossref","first-page":"5036","DOI":"10.21437\/Interspeech.2020-3015","article-title":"Conformer: Convolution-augmented Transformer for Speech Recognition","volume-title":"Proceedings of Interspeech 2020","author":"Gulati","year":"2020"},{"key":"2026061709271310100_bib39","doi-asserted-by":"crossref","first-page":"198","DOI":"10.1007\/978-3-319-99579-3_21","article-title":"TED-LIUM 3: Twice as Much Data and Corpus Repartition for Experiments on Speaker Adaptation","volume-title":"Speech and Computer","author":"Hernandez","year":"2018"},{"key":"2026061709271310100_bib40","doi-asserted-by":"crossref","first-page":"3099","DOI":"10.1121\/1.411872","article-title":"Acoustic characteristics of American English vowels","volume":"97","author":"Hillenbrand","year":"1995","journal-title":"The Journal of the Acoustical Society of America"},{"key":"2026061709271310100_bib41","doi-asserted-by":"crossref","first-page":"13","DOI":"10.1007\/978-3-031-04083-2_2","article-title":"Explainable AI Methods - A Brief Overview","volume-title":"xxAI - Beyond Explainable AI: International Workshop, Held in Conjunction with ICML 2020","author":"Holzinger","year":"2022"},{"key":"2026061709271310100_bib42","first-page":"8229","article-title":"Europarl-ST: A Multilingual Corpus for Speech Translation of Parliamentary Debates","volume-title":"Proceedings of 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","author":"Iranzo-S\u00e1nchez","year":"2020"},{"key":"2026061709271310100_bib43","doi-asserted-by":"crossref","first-page":"228","DOI":"10.1016\/j.patrec.2021.06.030","article-title":"Perturbation-based methods for explaining deep neural networks: A survey","volume":"150","author":"Ivanovs","year":"2021","journal-title":"Pattern Recognition Letters"},{"key":"2026061709271310100_bib44","first-page":"4198","article-title":"Towards Faithfully Interpretable NLP Systems: How Should We Define and Evaluate Faithfulness?","author":"Jacovi","year":"2020"},{"key":"2026061709271310100_bib45","doi-asserted-by":"crossref","first-page":"45","DOI":"10.18653\/v1\/2024.gebnlp-1.4","article-title":"An Explainable Approach to Understanding Gender Stereotype Text","volume-title":"Proceedings of the 5th Workshop on Gender Bias in Natural Language Processing (GeBNLP)","author":"Jeyaraj","year":"2024"},{"issue":"3 Pt 1","key":"2026061709271310100_bib46","doi-asserted-by":"crossref","first-page":"1252","DOI":"10.1121\/1.1288413","article-title":"Acoustic characteristics of English fricatives","volume":"108","author":"Jongman","year":"2000","journal-title":"The Journal of the Acoustical Society of America"},{"issue":"3","key":"2026061709271310100_bib47","doi-asserted-by":"crossref","first-page":"620","DOI":"10.3390\/ai4030033","article-title":"Explainable Image Classification: The Journey So Far and the Road Ahead","volume":"4","author":"Kamakshi","year":"2023","journal-title":"AI"},{"key":"2026061709271310100_bib48","doi-asserted-by":"crossref","first-page":"1639","DOI":"10.21437\/Interspeech.2020-2637","article-title":"Identifying important time-frequency locations in continuous speech utterances","volume-title":"Interspeech 2020","author":"Kavaki","year":"2020"},{"key":"2026061709271310100_bib49","volume-title":"Improving Feature Attribution through Input-specific Network Pruning","author":"Khakzar","year":"2020"},{"key":"2026061709271310100_bib50","article-title":"Adam: A Method for Stochastic Optimization","volume-title":"Proceedings of the 3rd International Conference on Learning Representations","author":"Kingma","year":"2015"},{"key":"2026061709271310100_bib51","first-page":"66","article-title":"SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing","volume-title":"Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing: System Demonstrations","author":"Kudo","year":"2018"},{"issue":"1","key":"2026061709271310100_bib52","doi-asserted-by":"crossref","first-page":"79","DOI":"10.1214\/aoms\/1177729694","article-title":"On Information and Sufficiency","volume":"22","author":"Kullback","year":"1951","journal-title":"The Annals of Mathematical Statistics"},{"issue":"S1","key":"2026061709271310100_bib53","doi-asserted-by":"crossref","first-page":"S80","DOI":"10.1121\/1.2029172","article-title":"Acoustic properties for place of articulation in nasal consonants: A cross-linguistic study","volume":"88","author":"Kurowski","year":"2005","journal-title":"The Journal of the Acoustical Society of America"},{"key":"2026061709271310100_bib54","volume-title":"Sparks of Large Audio Models: A Survey and Outlook","author":"Latif","year":"2023"},{"key":"2026061709271310100_bib55","first-page":"1630","article-title":"M4: A Unified XAI Benchmark for Faithfulness Evaluation of Feature Attribution Methods across Metrics, Modalities and Models","volume":"36","author":"Li","year":"2023","journal-title":"Advances on Neural Information Processing Systems"},{"key":"2026061709271310100_bib56","first-page":"4768","article-title":"A Unified Approach to Interpreting Model Predictions","volume-title":"Advances on Neural Information Processing Systems","author":"Lundberg","year":"2017"},{"issue":"8","key":"2026061709271310100_bib57","doi-asserted-by":"crossref","DOI":"10.1145\/3546577","article-title":"Post-hoc Interpretability for Neural NLP: A Survey","volume":"55","author":"Madsen","year":"2022","journal-title":"ACM Computing Surveys"},{"key":"2026061709271310100_bib58","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1109\/WASPAA.2013.6701835","article-title":"Learning an intelligibility map of individual utterances","volume-title":"2013 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics","author":"Mandel","year":"2013"},{"key":"2026061709271310100_bib59","doi-asserted-by":"crossref","first-page":"660","DOI":"10.21437\/Interspeech.2016-932","article-title":"Directly Comparing the Listening Strategies of Humans and Machines","volume-title":"Interspeech 2016","author":"Mandel","year":"2016"},{"issue":"6","key":"2026061709271310100_bib60","doi-asserted-by":"crossref","first-page":"3962","DOI":"10.1121\/1.2990715","article-title":"Acoustic characteristics of clearly spoken English fricatives","volume":"125","author":"Maniwa","year":"2009","journal-title":"The Journal of the Acoustical Society of America"},{"key":"2026061709271310100_bib61","first-page":"14","article-title":"Visualizing Automatic Speech Recognition \u2013 Means for a Better Understanding?","author":"Markert","year":"2021"},{"key":"2026061709271310100_bib62","article-title":"Locating and editing factual associations in GPT","volume-title":"Advances on Neural Information Processing Systems","author":"Meng","year":"2022"},{"key":"2026061709271310100_bib63","doi-asserted-by":"crossref","first-page":"165","DOI":"10.18653\/v1\/2023.nlposs-1.19","article-title":"Using Captum to Explain Generative Language Models","volume-title":"Proceedings of the 3rd Workshop for Natural Language Processing Open Source Software (NLP-OSS 2023)","author":"Miglani","year":"2023"},{"key":"2026061709271310100_bib64","doi-asserted-by":"crossref","first-page":"3378","DOI":"10.18653\/v1\/2023.eacl-main.245","article-title":"Quantifying Context Mixing in Transformers","volume-title":"Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics","author":"Mohebbi","year":"2023"},{"key":"2026061709271310100_bib65","doi-asserted-by":"crossref","first-page":"193","DOI":"10.1007\/978-3-030-28954-6_10","article-title":"Layer-Wise Relevance Propagation: An Overview","volume-title":"Explainable AI: Interpreting, Explaining and Visualizing Deep Learning","author":"Montavon","year":"2019"},{"key":"2026061709271310100_bib66","first-page":"120","article-title":"Explainable AI Discloses Gender Bias in Sexism Detection Algorithm","volume-title":"Proceedings of the 11th International Conference on Networking, Systems, and Security","author":"Muntasir","year":"2025"},{"key":"2026061709271310100_bib67","first-page":"2459","article-title":"We need to talk: HCI and the delicate topic of spoken language interaction","volume-title":"CHI \u201813 Extended Abstracts on Human Factors in Computing Systems","author":"Munteanu","year":"2013"},{"issue":"13s","key":"2026061709271310100_bib68","doi-asserted-by":"crossref","DOI":"10.1145\/3583558","article-title":"From Anecdotal Evidence to Quantitative Evaluation Methods: A Systematic Review on Evaluating Explainable AI","volume":"55","author":"Nauta","year":"2023","journal-title":"ACM Computing Surveys"},{"key":"2026061709271310100_bib69","first-page":"205","article-title":"Word Order Typology through Multilingual Word Alignment","volume-title":"Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)","author":"\u00d6stling","year":"2015"},{"key":"2026061709271310100_bib70","first-page":"5206","article-title":"Librispeech: An ASR corpus based on public domain audio books","volume-title":"Proceedings of 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","author":"Panayotov","year":"2015"},{"key":"2026061709271310100_bib71","doi-asserted-by":"crossref","first-page":"1139","DOI":"10.1145\/3593013.3594069","article-title":"The role of explainable AI in the context of the AI Act","volume-title":"Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency","author":"Panigutti","year":"2023"},{"key":"2026061709271310100_bib72","doi-asserted-by":"crossref","DOI":"10.21437\/Interspeech.2019-2680","article-title":"SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition","volume-title":"Proceedings of Interspeech 2019","author":"Park","year":"2019"},{"key":"2026061709271310100_bib73","first-page":"2221","article-title":"Explaining Speech Classification Models via Word-Level Audio Segments and Paralinguistic Features","volume-title":"Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","author":"Pastor","year":"2024"},{"key":"2026061709271310100_bib74","doi-asserted-by":"crossref","DOI":"10.1109\/ASRU57964.2023.10389676","article-title":"Reproducing Whisper-Style Training Using An Open-Source Toolkit And Publicly Available Data","volume-title":"2023 IEEE Automatic Speech Recognition and Understanding Workshop, ASRU 2023","author":"Peng","year":"2023"},{"key":"2026061709271310100_bib75","article-title":"RISE: Randomized Input Sampling for Explanation of Black-box Models","volume-title":"Proceedings of the British Machine Vision Conference (BMVC)","author":"Petsiuk","year":"2018"},{"key":"2026061709271310100_bib76","doi-asserted-by":"crossref","first-page":"247","DOI":"10.1145\/3514221.3517886","article-title":"Interpretable Data-Based Explanations for Fairness Debugging","volume-title":"Proceedings of the 2022 International Conference on Management of Data (SIGMOD)","author":"Pradhan","year":"2022"},{"issue":"3","key":"2026061709271310100_bib77","doi-asserted-by":"crossref","first-page":"225","DOI":"10.1016\/j.specom.2004.06.001","article-title":"Acoustic parameters for automatic detection of nasal manner","volume":"43","author":"Pruthi","year":"2004","journal-title":"Speech Communication"},{"key":"2026061709271310100_bib78","first-page":"28492","article-title":"Robust Speech Recognition via Large-Scale Weak Supervision","volume":"202","author":"Radford","year":"2023","journal-title":"Proceedings of the 40th International Conference on Machine Learning"},{"key":"2026061709271310100_bib79","first-page":"1135","article-title":"\u201cWhy Should I Trust You?\u201d: Explaining the Predictions of Any Classifier","volume-title":"Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD)","author":"Ribeiro","year":"2016"},{"issue":"3","key":"2026061709271310100_bib80","doi-asserted-by":"crossref","first-page":"1275","DOI":"10.1093\/ia\/iiae073","article-title":"Global AI governance: barriers and pathways forward","volume":"100","author":"Roberts","year":"2024","journal-title":"International Affairs"},{"key":"2026061709271310100_bib81","doi-asserted-by":"crossref","first-page":"11709","DOI":"10.18653\/v1\/2024.findings-emnlp.685","article-title":"A Comprehensive Survey of Hallucination in Large Language, Image, Video and Audio Foundation Models","volume-title":"Findings of the Association for Computational Linguistics: EMNLP 2024","author":"Sahoo","year":"2024"},{"issue":"11","key":"2026061709271310100_bib82","doi-asserted-by":"crossref","first-page":"2660","DOI":"10.1109\/TNNLS.2016.2599820","article-title":"Evaluating the Visualization of What a Deep Neural Network Has Learned","volume":"28","author":"Samek","year":"2017","journal-title":"IEEE Transactions on Neural Networks and Learning Systems"},{"issue":"3","key":"2026061709271310100_bib83","doi-asserted-by":"crossref","first-page":"247","DOI":"10.1109\/JPROC.2021.3060483","article-title":"Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications","volume":"109","author":"Samek","year":"2021","journal-title":"Proceedings of the IEEE"},{"key":"2026061709271310100_bib84","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1007\/978-3-030-28954-6_1","article-title":"Towards Explainable Artificial Intelligence","volume-title":"Explainable AI: Interpreting, Explaining and Visualizing Deep Learning","author":"Samek","year":"2019"},{"key":"2026061709271310100_bib85","doi-asserted-by":"crossref","first-page":"421","DOI":"10.18653\/v1\/2023.acl-demo.40","article-title":"Inseq: An Interpretability Toolkit for Sequence Generation Models","volume-title":"Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)","author":"Sarti","year":"2023"},{"key":"2026061709271310100_bib86","doi-asserted-by":"crossref","first-page":"240","DOI":"10.18653\/v1\/2021.blackboxnlp-1.17","article-title":"Efficient Explanations from Empirical Explainers","volume-title":"Proceedings of the Fourth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP","author":"Schwarzenberg","year":"2021"},{"key":"2026061709271310100_bib87","first-page":"618","article-title":"Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV)","author":"Selvaraju","year":"2019"},{"key":"2026061709271310100_bib88","doi-asserted-by":"crossref","first-page":"1715","DOI":"10.18653\/v1\/P16-1162","article-title":"Neural Machine Translation of Rare Words with Subword Units","volume-title":"Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","author":"Sennrich","year":"2016"},{"key":"2026061709271310100_bib89","doi-asserted-by":"crossref","first-page":"8572","DOI":"10.1109\/ACCESS.2019.2963055","article-title":"Regional Multi-Scale Approach for Visually Pleasing Explanations of Deep Neural Networks","volume":"8","author":"Seo","year":"2020","journal-title":"IEEE Access"},{"key":"2026061709271310100_bib90","first-page":"3145","article-title":"Learning Important Features Through Propagating Activation Differences","volume":"70","author":"Shrikumar","year":"2017","journal-title":"Proceedings of the 34th International Conference on Machine Learning"},{"key":"2026061709271310100_bib91","article-title":"Deep inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps","volume-title":"Proceedings of the 2nd International Conference on Learning Representations","author":"Simonyan","year":"2014"},{"key":"2026061709271310100_bib92","volume-title":"Acoustic Phonetics","author":"Stevens","year":"2000"},{"key":"2026061709271310100_bib93","doi-asserted-by":"crossref","first-page":"391","DOI":"10.1007\/978-3-030-58604-1_24","article-title":"Explaining Image Classifiers Using Statistical Fault Localization","volume-title":"Computer Vision\u2013ECCV 2020: 16th European Conference, Proceedings","author":"Sun","year":"2020"},{"key":"2026061709271310100_bib94","first-page":"3319","article-title":"Axiomatic Attribution for Deep Networks","volume-title":"Proceedings of the 34th International Conference on Machine Learning","author":"Sundararajan","year":"2017"},{"key":"2026061709271310100_bib95","article-title":"Sequence to Sequence Learning with Neural Networks","volume-title":"Advances on Neural Information Processing Systems","author":"Sutskever","year":"2014"},{"key":"2026061709271310100_bib96","first-page":"2818","article-title":"Rethinking the Inception Architecture for Computer Vision","volume-title":"Proceedings of the 29th IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Szegedy","year":"2016"},{"issue":"04","key":"2026061709271310100_bib97","doi-asserted-by":"crossref","first-page":"6021","DOI":"10.1609\/aaai.v34i04.6064","article-title":"Sanity Checks for Saliency Metrics","volume":"34","author":"Tomsett","year":"2020","journal-title":"Proceedings of the AAAI Conference on Artificial Intelligence"},{"key":"2026061709271310100_bib98","doi-asserted-by":"crossref","first-page":"312","DOI":"10.1109\/TASLP.2020.3040545","article-title":"Directly Comparing the Listening Strategies of Humans and Machines","volume":"29","author":"Trinh","year":"2021","journal-title":"IEEE\/ACM Transactions on Audio, Speech, and Language Processing"},{"key":"2026061709271310100_bib99","doi-asserted-by":"crossref","first-page":"1166","DOI":"10.21437\/Interspeech.2020-2883","article-title":"Large Scale Evaluation of Importance Maps in Automatic Speech Recognition","volume-title":"Interspeech 2020","author":"Trinh","year":"2020"},{"key":"2026061709271310100_bib100","doi-asserted-by":"crossref","first-page":"1616","DOI":"10.21437\/Interspeech.2018-2377","article-title":"Bubble Cooperative Networks for Identifying Important Speech Cues","volume-title":"Interspeech 2018","author":"Trinh","year":"2018"},{"key":"2026061709271310100_bib101","article-title":"Attention is All you Need","volume-title":"Advances in Neural Information Processing Systems","author":"Vaswani","year":"2017"},{"issue":"1","key":"2026061709271310100_bib102","doi-asserted-by":"crossref","first-page":"133","DOI":"10.1016\/S0167-6393(98)00033-8","article-title":"Cepstral domain segmental feature vector normalization for noise robust speech recognition","volume":"25","author":"Viikki","year":"1998","journal-title":"Speech Communication"},{"key":"2026061709271310100_bib103","doi-asserted-by":"crossref","first-page":"89","DOI":"10.1016\/j.inffus.2021.05.009","article-title":"Notions of explainability and evaluation approaches for explainable artificial intelligence","volume":"76","author":"Vilone","year":"2021","journal-title":"Information Fusion"},{"key":"2026061709271310100_bib104","first-page":"9089","article-title":"Interpretable and Fine-Grained Visual Explanations for Convolutional Neural Networks","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Wagner","year":"2019"},{"key":"2026061709271310100_bib105","first-page":"4197","article-title":"CoVoST: A Diverse Multilingual Speech-To-Text Translation Corpus","volume-title":"Proceedings of the Twelfth Language Resources and Evaluation Conference","author":"Wang","year":"2020"},{"key":"2026061709271310100_bib106","first-page":"993","article-title":"VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation Learning, Semi-Supervised Learning and Interpretation","volume-title":"Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)","author":"Wang","year":"2021"},{"key":"2026061709271310100_bib107","first-page":"33","article-title":"Fairseq S2T: Fast Speech-to-Text Modeling with Fairseq","volume-title":"Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing: System Demonstrations","author":"Wang","year":"2020"},{"key":"2026061709271310100_bib108","volume-title":"Ethical and social risks of harm from Language Models","author":"Weidinger","year":"2021"},{"key":"2026061709271310100_bib109","doi-asserted-by":"crossref","DOI":"10.1109\/ICASSP49357.2023.10094635","article-title":"Explanations for Automatic Speech Recognition","volume-title":"Proceedings of 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","author":"Wu","year":"2023"},{"key":"2026061709271310100_bib110","doi-asserted-by":"crossref","first-page":"10296","DOI":"10.1109\/ICASSP48485.2024.10445989","article-title":"Can We Trust Explainable AI Methods on ASR? An Evaluation on Phoneme Recognition","volume-title":"ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","author":"Wu","year":"2024"},{"key":"2026061709271310100_bib111","volume-title":"Efficient Streaming Language Models with Attention Sinks","author":"Xiao","year":"2024"},{"key":"2026061709271310100_bib112","first-page":"1376","article-title":"MFPP: Morphological Fragmental Perturbation Pyramid for Black-Box Model Explanations","volume-title":"Proceedings of the 25th International Conference on Pattern Recognition (ICPR)","author":"Yang","year":"2021"},{"key":"2026061709271310100_bib113","unstructured":"Jinsung\n              Yoon\n            , JamesJordon, and Mihaelavan der Schaar. 2019. INVASE: Instance-wise Variable Selection using Neural Networks. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6\u20139, 2019. OpenReview.net."},{"key":"2026061709271310100_bib114","doi-asserted-by":"crossref","first-page":"818","DOI":"10.1007\/978-3-319-10590-1_53","article-title":"Visualizing and Understanding Convolutional Networks","volume-title":"Computer Vision \u2013 ECCV 2014 13th European Conference, Proceedings","author":"Zeiler","year":"2014"},{"key":"2026061709271310100_bib115","volume-title":"Towards Best Practices of Activation Patching in Language Models: Metrics and Methods","author":"Zhang","year":"2024"},{"issue":"2","key":"2026061709271310100_bib116","doi-asserted-by":"crossref","DOI":"10.1145\/3639372","article-title":"Explainability for Large Language Models: A Survey","volume":"15","author":"Zhao","year":"2024","journal-title":"ACM Transactions on Intelligent Systems and Technology"},{"issue":"3","key":"2026061709271310100_bib117","doi-asserted-by":"crossref","first-page":"464","DOI":"10.1109\/TAI.2022.3207982","article-title":"Sparsing and Smoothing for the seq2seq Models","volume":"4","author":"Zhao","year":"2023","journal-title":"IEEE Transactions on Artificial Intelligence"}],"container-title":["Transactions of the Association for Computational Linguistics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/TACL.a.684\/2606986\/tacl.a.684.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/TACL.a.684\/2606986\/tacl.a.684.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,17]],"date-time":"2026-06-17T13:27:43Z","timestamp":1781702863000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/tacl\/article\/doi\/10.1162\/TACL.a.684\/137205\/SPES-Spectrogram-Perturbation-for-Explainable"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":117,"URL":"https:\/\/doi.org\/10.1162\/tacl.a.684","relation":{},"ISSN":["2307-387X"],"issn-type":[{"value":"2307-387X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026]]},"published":{"date-parts":[[2026]]}}}