Automated Legal Document Drafting in the Indian Judicial System: A Survey of AI-driven Approaches

Abstract

The Indian judicial system faces many challenges which slow down legal document preparation because of the complex procedures that is present in judicial system, numerous cases and language differences. The process of manual tran- scription of different languages, extraction of entities and court standard compliance formatting needs help of lawyers to spend a lot of time which results in late and incorrect work. The research states that we have done evaluation of 20 academic papers from 2017 to 2025 which aims to investigate Automatic Speech Recognition (ASR), Natural Language Processing (NLP) and transfer learning models for legal systems. The research shows that Whisper ASR models achieve a high level performance with help of 55.2% error reduction from previous models but these improvements focus on Western regions that use languages with a lot of resources. Our current state-of-the-art models show very low applicability for Indian legal systems because they were developed for Western jurisdictions with high-resource languages. The current systems does not have essential features because they do not handle issues like overlapping speech and they do not integrate multiple techniques like multilingual transcription, speaker diarization, legal entity extraction and jurisdiction- aware template generation. The Indian judicial system requires solutions to these problems because they create problems for the public access to justice and prevent judicial advancements.

Introduction

Legal document preparation is a critical yet labor-intensive requirement of judicial systems worldwide. In India, this process is particularly challenging due to multilingual court proceedings, diverse jurisdictions, and reliance on manual note-taking and transcription, which lead to human errors, delays, and a growing backlog of unresolved cases. Automating the conversion of spoken courtroom dialogue into structured legal documents using Artificial Intelligence can significantly reduce judicial workload and improve accessibility and efficiency.

The integration of Automatic Speech Recognition (ASR), speaker diarization, and legal domain–specific Natural Language Processing (NLP) has the potential to transform judicial operations. However, existing systems face major limitations, including poor recognition of legal terminology, difficulty distinguishing overlapping speakers, weak performance in noisy courtroom environments, and limited support for Indian languages. These challenges highlight the need for customized, integrated, and scalable solutions tailored to the Indian judicial context.

This survey systematically reviews current research on automated legal document generation, focusing on four core technical components: multilingual ASR, speaker diarization, legal-specific NLP, and parameter-efficient model optimization. The literature is categorized into key areas such as ASR and diarization, legal NLP and transfer learning, dataset creation under resource scarcity, model efficiency, and integrated system design. While individual components have advanced significantly, their lack of integration and domain adaptation limits real-world deployment.

Findings reveal substantial research gaps, particularly the scarcity of high-quality annotated Indian legal datasets, insufficient multilingual coverage, limited handling of overlapping speech, and the absence of end-to-end systems that generate jurisdiction-compliant legal documents directly from audio. Although models like Whisper, Legal-BERT, and Indic-transformers show promise, they require domain adaptation, efficient fine-tuning techniques such as LoRA, and larger multilingual datasets to be effective in Indian courts.

Overall, the study concludes that while AI technologies for ASR and legal NLP have matured independently, the Indian judicial system lacks a fully integrated, ethical, and scalable automation framework. Future research must prioritize multilingual dataset expansion, robust speaker diarization, legal-aware document structuring, and the development of end-to-end systems capable of converting courtroom speech into standardized legal documents compliant with Indian jurisdictional requirements.

Conclusion

The survey provides a complete assessment of AI-based systems which generate legal documents automatically for the Indian judicial system. The evaluation of 36 papers spanning from 2017 to 2025 shows major progress in ASR technology because Whisper achieved a 55.2% error decrease and Legal- BERT delivered improved results for legal NLP applications in specific domains. The current technology fails to provide solutions that meet the particular needs of the Indian judicial system. The research shows that Indian language datasets have insufficient data and speaker diarization systems fail to work properly with noisy audio because of legal restrictions against complete end-to-end automation. The research identifies three main gaps in current studies. The surveys and datasets fail to support enough Indian languages because they only cover 33.3% of surveys and 28.6% datasets address low- resource languages. The reviewed work lacks complete audio- to-document pipelines because no study presents integrated systems. The current systems fail to adapt their legal terminol- ogy and procedures to the specific requirements of the Indian legal domain. The research must fulfill three essential goals which in- clude (1) creating professionally identified datasets for In- dian legal procedures and languages through IL-TUR and IndianBailJudgments-1200 models; (2) The system will un- dergo ethical and fairness assessments through state-of-the-art bias detection and model interpretability tools which preserve legal responsibility; (3) The system needs workflow integration to perform automated e-filing through direct access to the E- Court System; (4) The system needs scalability improvement through lightweight fine-tuning methods which work well for courts with restricted funding. The development of integrated systems which merge speech recognition with NLP and jurisdiction-specific components will establish efficient legal documentation solutions that work for all linguistic groups in India.

References

[1] Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey and I. Sutskever, ”Robust speech recognition via large-scale weak supervision,” in Proc. Int. Conf. on Machine Learning (ICML), pp. 28492–28518, PMLR, July 2023. [2] A. Lyra, C. E. Barbosa, H. Salazar, M. Argoˆlo, Y. Lima, R. Motta and J. M. de Souza, ”Automatic Transcription Systems: A Game Changer for Court Hearings,” 2024. [3] A. Conneau, et al., ”Fleurs: Few-shot learning evaluation of universal representations of speech,” in 2022 IEEE Spoken Language Technology Workshop (SLT), pp. 798–805, IEEE, January 2023. [4] I. Chalkidis, M. Fergadiotis, P. Malakasiotis, N. Aletras and I. Androut- sopoulos, ”LEGAL-BERT: The muppets straight out of law school,” arXiv preprint arXiv:2010.02559, 2020. [5] D. Garcia-Romero, D. Snyder, G. Sell, D. Povey and A. McCree, ”Speaker diarization using deep neural network embeddings,” in 2017 IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP), pp. 4930–4934, IEEE, March 2017. [6] I. Chalkidis, I. Androutsopoulos and N. Aletras, ”Neural legal judgment prediction in English,” arXiv preprint arXiv:1906.02059, 2019. [7] K. Jain, A. Deshpande, K. Shridhar, F. Laumann and A. Dash, ”Indic- transformers: An analysis of transformer language models for indian languages,” arXiv preprint arXiv:2011.02323, 2020. [8] E. J. Hu, et al., ”Lora: Low-rank adaptation of large language models,” in ICLR, vol. 1, no. 2, p. 3, 2022. [9] K. Ding, et al., ”A simple and effective method to improve zero-shot cross-lingual transfer learning,” arXiv preprint arXiv:2210.09934, 2022. [10] C. Raffel, et al., ”Exploring the limits of transfer learning with a unified text-to-text transformer,” in Journal of Machine Learning Research, vol. 21, no. 140, pp. 1–67, 2020. [11] A. Joshi, S. Paul, A. Sharma, P. Goyal, S. Ghosh and A. Modi, ”IL- TUR: Benchmark for Indian legal text understanding and reasoning,” arXiv preprint arXiv:2407.05399, 2024. [12] S. Deshmukh and P. Kamble, ”IndianBailJudgments-1200: A Multi- Attribute Dataset for Legal NLP on Indian Bail Orders,” arXiv preprint arXiv:2507.02506, 2025. [13] R. Barman, ”Unveiling the Future: The Intersection of Artificial Intelli- gence and the Judicial System,” in Indian J. Integrated Rsch. L., vol. 3, p. 1, 2023. [14] R. S. D. Oliveira and E. G. Sperandio Nascimento, ”Analysing similar- ities between legal court documents using natural language processing approaches based on transformers,” in PloS One, vol. 20, no. 4, e0320244, 2025. [15] K. Shreeram, A. Patil, A. Agarwal, D. P. Rana and R. G. Mehta, ”Natural Language Understanding (NLU)-Powered Information Retrieval from Indian Legal Property Graph,” in Procedia Computer Science, vol. 258, pp. 2500–2509, 2025. [16] M. M. Rahman, N. M. Gony, M. M. Rahman, M. Rahman and S. D. Khatun Shuvra, ”Natural language processing in legal document analysis software: A systematic review of current approaches, challenges, and opportunities,” in Int. Journal of Innovative Research and Scientific Studies, vol. 8, no. 3, pp. 5026–5042, 2025. [17] S. Castano, et al., ”Enforcing legal information extraction through context-aware techniques: The ASKE approach,” in Computer Law & Security Review, vol. 52, 105903, 2024. [18] S. Adhikary, D. Roy, D. Ganguly, S. Kumar Guha and K. Ghosh, ”Leda: a system for legal data annotation,” in Frontiers in Artificial Intelligence and Applications, no. 379, pp. 370–367, 2023. [19] D. H. Anh, D. T. Do, V. Tran and N. Le Minh, ”The impact of large language modeling on natural language processing in legal texts: A comprehensive survey,” in 2023 15th Int. Conf. on Knowledge and Systems Engineering (KSE), pp. 1–7, IEEE, October 2023. [20] A. Farzindar and G. Lapalme, ”Letsum, an automatic legal text summa- rizing,” in Legal knowledge and information systems: JURIX, vol. 11, 2004. [21] P. Krasadakis, E. Sakkopoulos and V. S. Verykios, ”A survey on challenges and advances in natural language processing with a focus on legal informatics and low-resource languages,” in Electronics, vol. 13, no. 3, p. 648, 2024. [22] M. Yang, et al., ”Low-rank adaptation for foundation models: A com- prehensive review,” arXiv preprint arXiv:2501.00365, 2024. [23] P. Kalamkar, A. Agarwal, A. Tiwari, S. Gupta, S. Karn and V. Raghavan, ”Named entity recognition in Indian court judgments,” arXiv preprint arXiv:2211.03442, 2022. [24] I. Chalkidis, M. Fergadiotis and I. Androutsopoulos, ”MultiEURLEX– A multi-lingual and multi-label legal document classification dataset for zero-shot cross-lingual transfer,” arXiv preprint arXiv:2109.00904, 2021. [25] R. M. R. Kadiyala, S. Pullakhandam, K. Mehreen, S. Tippareddy and A. Srivastava, ”Augmenting Legal Decision Support Systems with LLM-based NLI for Analyzing Social Media Evidence,” arXiv preprint arXiv:2410.15990, 2024. [26] T. Y. S. S. Santosh, K. D. Ashley, K. Atkinson and M. Grabmair, ”Towards supporting legal argumentation with NLP: Is more data really all you need?,” in Proc. of the Natural Legal Language Processing Workshop 2024, pp. 404–421, November 2024. [27] F. Ariai, J. Mackenzie and G. Demartini, ”Natural language processing for the legal domain: A survey of tasks, datasets, models, and chal- lenges,” arXiv preprint arXiv:2410.21306, 2024. [28] R. Mahari, D. Stammbach, E. Ash and A. Pentland, ”The law and NLP: Bridging disciplinary disconnects,” in Findings of the Association for Computational Linguistics: EMNLP 2023, pp. 3445–3454, December 2023. [29] R. Christen, A. Shaitarova, M. Stu¨rmer and J. Niklaus, ”Resolving legalese: A multilingual exploration of negation scope resolution in legal documents,” in Proc. of the 2024 Joint Int. Conf. on Computational Lin- guistics, Language Resources and Evaluation (LREC-COLING 2024), pp. 13992–14004, May 2024. [30] A. Pahilajani, S. Jain and D. Trivedi, ”NLP at UC Santa Cruz at SemEval-2024 task 5: Legal answer validation using few-shot multi- choice QA,” in Proc. of the 18th Int. Workshop on Semantic Evaluation (SemEval-2024), pp. 1309–1314, June 2024. [31] J. Li, R. Bhambhoria, S. Dahan and X. Zhu, ”Experimenting with legal ai solutions: The case of question-answering for access to justice,” arXiv preprint arXiv:2409.07713, 2024. [32] D. Datta, S. Soni, R. Mukherjee and S. Ghosh, ”MILDSum: A novel benchmark dataset for multilingual summarization of Indian legal case judgments,” arXiv preprint arXiv:2310.18600, 2023. [33] K. Swamy, O. Salgare, O. Sakhare and S. Tamboli, ”ValidEase: NLP for Simplification and Summarization of Legal Documents,” 2024. [34] J. J. W. Ng, et al., ”Evaluating the performance of artificial intelligence- based speech recognition for clinical documentation: a systematic re- view,” in BMC Medical Informatics and Decision Making, vol. 25, no. 1, p. 236, 2025. [35] M. M. Rahman, N. Gony, M. M. Rahman, M. M. Rahman and M. SD, ”Natural language processing in legal document analysis: A systematic review of current approaches,” in Int. J. Innov. Res. Sci. Stud., vol. 8, pp. 5016–5072, 2025. [36] D. M. Katz, D. Hartung, L. Gerlach, A. Jana and M. J. Bommarito II, ”Natural language processing in the legal domain,” arXiv preprint arXiv:2302.12039, 2023.

Copyright

Copyright © 2026 Hemamalini S, Md Raqeeb Haider, Prajwal Jahagirdar, Shashwat Kumar, Dr. Rohini T V. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.