Listening to the archipelago: automatic speech recognition, accent variation, and the unequal modeling of Indonesian voices

Imam Masrur

doi:10.67490/ijcl.v1i3.1026

Authors

Imam Masrur Universitas Islam Negeri Syekh Wasil Kediri Author

DOI:

https://doi.org/10.67490/ijcl.v1i3.1026

Keywords:

accent variation, automatic speech recognition, Indonesian languages, sociotechnical visibility, speech technology

Abstract

Background: Indonesia’s archipelagic linguistic ecology makes automatic speech recognition (ASR) a problem of sociotechnical representation, because spoken Indonesian is shaped by regional accent, local-language influence, code-switching, and uneven access to standardized speech resources. Objective: This study examines how Indonesian ASR resources, evaluation practices, and model documentation make regional and accent-varied voices differentially visible, testable, and interpretable. Method: This study applies a corpus-based ASR audit to 18 public and verifiable corpus items, combining archipelagic representation analysis, accent-sensitive error-evaluation readiness, and sociotechnical visibility assessment. Results: The corpus shows that regional-language resources are more visible than dialectal Indonesian speech directly suitable for accent-sensitive ASR evaluation. Error-evaluation readiness is uneven because several items require transcript alignment, utterance segmentation, metadata enrichment, or model-output comparison before reliable subgroup analysis can be conducted. Sociotechnical visibility is fragmented across direct speech evidence, preprocessing-dependent materials, literature-mediated findings, and model-opaque documentation. Implication: These findings reposition Indonesian ASR fairness as a question of audibility, evaluability, and linguistic justice rather than aggregate transcription accuracy alone. Novelty: This study contributes an archipelagic framework for assessing unequal machine listening in Indonesian ASR by integrating corpus representation, evaluation readiness, and sociotechnical visibility.

References

[1] T. Abidin, A. Misbullah, R. Ferdhiana, L. Farsiah, M. Z. Aksana, and H. Riza, "Acoustic model with multiple lexicon types for Indonesian speech recognition," Applied Computational Intelligence and Soft Computing, vol. 2022, Art. 3227828, 2022, doi: 10.1155/2022/3227828.

[2] P. Arisaputra and A. Zahra, "Indonesian automatic speech recognition with XLSR-53," Ingénierie des Systèmes d’Information, vol. 27, no. 6, 2022, doi: 10.18280/isi.270614.

[3] P. Arisaputra, A. T. Handoyo, and A. Zahra, "XLS-R deep learning model for multilingual ASR on low-resource languages: Indonesian, Javanese, and Sundanese," arXiv, 2024, doi: 10.48550/arXiv.2401.06832.

[4] A. Adila, D. Lestari, A. Purwarianti, D. Tanaya, K. Azizah, and S. Sakti, "Enhancing Indonesian automatic speech recognition: Evaluating multilingual models with diverse speech variabilities," in Proc. 2024 27th Conf. Oriental COCOSDA, 2024, pp. 1-6, doi: 10.1109/O-COCOSDA64382.2024.10800336.

[5] R. V. M. Tazakka, D. Lestari, A. Purwarianti, D. Tanaya, K. Azizah, and S. Sakti, "Indonesian-English code-switching speech recognition using machine speech chain based semi-supervised learning," in Proc. 3rd Annual Meeting SIGUL @ LREC-COLING 2024, 2024, doi: 10.63317/3hriruv4bj9u.

[6] S. Feng, B. Halpern, O. Kudina, and O. Scharenborg, "Towards inclusive automatic speech recognition," Computer Speech & Language, vol. 84, Art. 101567, 2024, doi: 10.1016/j.csl.2023.101567.

[7] C. Graham and N. Roll, "Evaluating OpenAI’s Whisper ASR: Performance analysis across diverse accents and speaker traits," JASA Express Letters, vol. 4, no. 2, 2024, doi: 10.1121/10.0024876.

[8] N. Markl, "Language variation and algorithmic bias: Understanding algorithmic bias in British English automatic speech recognition," in Proc. 2022 ACM Conf. Fairness, Accountability, and Transparency, 2022, doi: 10.1145/3531146.3533117.

[9] A. Wassink, C. Gansen, and I. Bartholomew, "Uneven success: Automatic speech recognition and ethnicity-related dialects," Speech Communication, vol. 140, pp. 50-70, 2022, doi: 10.1016/j.specom.2022.03.009.

[10] T. Patel, W. Hutiri, A. Y. Ding, and O. Scharenborg, "How to evaluate automatic speech recognition: Comparing different performance and bias measures," arXiv, 2025, doi: 10.48550/arXiv.2507.05885.

[11] K. Prinos, N. Patwari, and C. A. Power, "Speaking of accent: A content analysis of accent misconceptions in ASR research," in Proc. 2024 ACM Conf. Fairness, Accountability, and Transparency, 2024, doi: 10.1145/3630106.3658969.

[12] K. Koźmińska, "Reimagining accents and speech recognition with sociolinguistic perception studies and research on listening subjects," Journal of Sociolinguistics, 2025, doi: 10.1111/josl.12720.

[13] A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, "Robust speech recognition via large-scale weak supervision," in Proc. 40th Int. Conf. Machine Learning, 2022, pp. 28492-28518.

[14] H. Kheddar, M. Hemis, and Y. Himeur, "Automatic speech recognition using advanced deep learning approaches: A survey," Information Fusion, 2024, doi: 10.1016/j.inffus.2024.102422.

[15] A. Jarin, A. Santosa, M. T. Uliniansyah, L. R. Aini, E. Nurfadhilah, and G. Gunarso, "Automatic speech recognition for Indonesian medical dictation in cloud environment," IAES International Journal of Artificial Intelligence, 2024, doi: 10.11591/ijai.v13.i2.pp1762-1772.

[16] H. Pardede, P. Adhi, V. Zilvan, A. Ramdan, and D. Krisnandi, "Deep convolutional neural networks-based features for Indonesian large vocabulary speech recognition," IAES International Journal of Artificial Intelligence, vol. 12, no. 2, pp. 610-617, 2023, doi: 10.11591/ijai.v12.i2.pp610-617.

[17] D. Dwi, P. D. Pamungkasari, A. P. Rifai, I. D. Sholihati, and G. Sakarkar, "Comparison of feature extraction techniques for long short-term memory models in Indonesian automatic speech recognition," Green Intelligent Systems and Applications, 2025, doi: 10.53623/gisa.v5i1.605.

[18] H. Azis, M. F. F. M, H. Darwis, P., T. Hasanuddin, and S., "Development of a low-resource automatic speech recognition system for the Makassar dialect," in Proc. 2026 20th Int. Conf. Ubiquitous Information Management and Communication (IMCOM), 2026, pp. 1-8, doi: 10.1109/IMCOM69009.2026.11360893.

[19] K. Tomanek, V. Zayats, D. Padfield, K. Vaillancourt, and F. Biadsy, "Residual adapters for parameter-efficient ASR adaptation to atypical and accented speech," in Proc. 2021 Conf. Empirical Methods in Natural Language Processing, 2021, doi: 10.18653/v1/2021.emnlp-main.541.

[20] O. H. Anidjar, R. Marbel, and R. Yozevitch, "Whisper turns stronger: Augmenting Wav2Vec 2.0 for superior ASR in low-resource languages," arXiv, 2024, doi: 10.48550/arXiv.2501.00425.

[21] A. Hadziqi, "Small models, many languages: Parameter-efficient adaptation of multilingual language models for low-resource Indonesian languages," Indonesian Journal of Computational Language Studies, vol. 1, no. 2, 2026.

[22] G. Brown and S. Hellmuth, "Computational modelling of segmental and prosodic levels of analysis for capturing variation across Arabic dialects," Speech Communication, vol. 141, pp. 80-92, 2022, doi: 10.1016/j.specom.2022.05.003.

[23] M. Del Rio, C. Miller, J. Profant, J. Drexler-Fox, Q. McNamara, N. Bhandari, N. Delworth, I. Pirkin, M. Jette, S. Chandra, P. Ha, and R. Westerman, "Accents in speech recognition through the lens of a World Englishes evaluation set," Research in Language, 2023, doi: 10.18778/1731-7533.21.3.02.

[24] L.-F. Lai, J. G. Van Hell, and J. Lipski, "Dialect bias in automatic speech recognition: Analysis of Appalachian English," American Speech: A Quarterly of Linguistic Usage, 2024, doi: 10.1215/00031283-11466494.

[25] D. Serditova and K. Tang, "Automatic speech recognition biases in Newcastle English: An error analysis," in Proc. Interspeech 2025, 2025, doi: 10.21437/interspeech.2025-1973.

[26] N. Das, S. Bodapati, M. Sunkara, S. Srinivasan, and D. H. Chau, "Best of both worlds: Robust accented speech recognition with adversarial transfer learning," in Proc. Interspeech 2021, 2021, pp. 1314-1318, doi: 10.21437/Interspeech.2021-1888.

[27] X. Wang, Y. Long, and D. Xu, "Universal and accent-discriminative encoders for conformer-based accent-invariant speech recognition," International Journal of Speech Technology, vol. 25, pp. 987-995, 2022, doi: 10.1007/s10772-022-10010-z.

[28] D. Prabhu, P. Jyothi, S. Ganapathy, and V. Unni, "Accented speech recognition with accent-specific codebooks," arXiv, 2023, doi: 10.48550/arXiv.2310.15970.

[29] T. Banerjee and V. Ramasubramanian, "Accent-robust speech recognition for English in low-resource settings using manifold mixup," EURASIP Journal on Audio, Speech, and Music Processing, vol. 2025, 2025, doi: 10.1186/s13636-025-00435-0.

[30] W. Lee, H. Kim, and G. Lee, "Mixture-of-experts with intermediate CTC supervision for accented speech recognition," arXiv, 2026, doi: 10.48550/arXiv.2602.01967.

[31] A. Bera and A. Agarwal, "Enhancing non-native accent recognition through a combination of speaker embeddings, prosodic and vocal speech features," Signal & Image Processing: An International Journal, 2023, doi: 10.5121/sipij.2023.14302.

[32] S. Ghorbani and J. H. L. Hansen, "Advanced accent/dialect identification and accentedness assessment with multi-embedding models and automatic speech recognition," The Journal of the Acoustical Society of America, vol. 155, no. 6, pp. 3848-3860, 2023, doi: 10.1121/10.0026235.

[33] M. A. Dar and J. Pushparaj, "Machine learning and deep learning approaches for accent recognition: A review," IEEE Access, vol. 13, pp. 51527-51550, 2025, doi: 10.1109/ACCESS.2025.3552935.

[34] S. Feng, O. Kudina, B. Halpern, and O. Scharenborg, "Quantifying bias in automatic speech recognition," arXiv, 2021.

[35] C. Li, T. Cohen, and S. V. S. Pakhomov, "Reexamining racial disparities in automatic speech recognition performance: The role of confounding by provenance," arXiv, 2024, doi: 10.48550/arXiv.2407.13982.

[36] A. DiChristofano, H. Shuster, S. Chandra, and N. Patwari, "Performance disparities between accents in automatic speech recognition," arXiv, 2022, doi: 10.48550/arXiv.2208.01157.

[37] M. Jahan, P. Mazumdar, T. Thebaud, M. Hasegawa-Johnson, J. Villalba, N. Dehak, and L. Moro-Velázquez, "Unveiling performance bias in ASR systems: A study on gender, age, accent, and more," in Proc. ICASSP 2025, 2025, pp. 1-5, doi: 10.1109/ICASSP49660.2025.10887767.

[38] A. Rai, S. D. Jaiswal, S. Prakash, B. P. Sree, and A. Mukherjee, "DENOASR: Debiasing ASRs through selective denoising," in Proc. 2024 IEEE Int. Conf. Knowledge Graph (ICKG), 2024, pp. 283-290, doi: 10.1109/ICKG63256.2024.00043.

[39] A. Martin, R. L. MacDonald, P. Jiang, M. A. Ladewig, J. Cattiau, R. Heywood, R. Cave, J. Tobin, P. C. Nelson, and K. Tomanek, "Project Euphonia: Advancing inclusive speech recognition through expanded data collection and evaluation," Frontiers in Language Sciences, 2025, doi: 10.3389/flang.2025.1569448.

[40] J.-B. Sun, Y. Xiao, S. Chung, Q. Hu, G. Huang, E.-J. Holden, and T. Dang, "Activation steering for accent adaptation in speech foundation models," 2026.

[41] M. A. Sumadi, "Who gets represented by Indonesian AI? Measuring regional, gender, and sociolinguistic bias in large language models," Indonesian Journal of Computational Language Studies, vol. 1, no. 2, 2026.