Sign in Subscribe

Research & Publications

Kathleen Siminyu's research centers on Natural Language Processing (NLP), automatic speech recognition (ASR), and data governance for low-resourced African languages. Her work treats language technology development as a form of socio-cultural activism, working to dismantle digital inequities and champion indigenous language representation on computational platforms.

Active Research Focus

Evaluating Speech Data Optimization and User Needs for Kiswahili Automatic Speech Recognition

MSc Research, University of Pretoria (2023 – Present / Concluding)

Building effective Automatic Speech Recognition (ASR) for African languages requires moving beyond one-size-fits-all, compute-heavy paradigms toward data efficiency, linguistic grounding, and real-world utility. This research investigates the full lifecycle of Kiswahili ASR systems across three core questions:

  • Data Optimization & Scarcity: How can we maximize the utility of existing speech recognition data to train high-performing Kiswahili models? (Examining data repetition limits before performance degradation and speech augmentation techniques).
  • Grounded User Needs: What are the practical shortcomings of existing speech recognition systems for content creators engaged in digital knowledge preservation? (Investigated through observational studies and focus group discussions [FGD] with creators and language stewards).
  • Linguistic Frameworks & Evaluation: How can African linguistic frameworks be leveraged to build better speech models and diagnostic evaluation methods? (Fine-tuning Massively Multilingual Speech [MMS] models and performing linguistically grounded error analyses).

Publications & Preprints

First-Authored Publications

  1. On the Tolerance of Repetition Before Performance Degradation in Kiswahili Automatic Speech Recognition
    • Kathleen Siminyu, Kathy Reid.
    • Proceedings of the 6th Workshop on African Natural Language Processing (AfricaNLP 2025), Vienna, Austria, 2025. ACL Anthology
  2. Consultative engagement of stakeholders toward a roadmap for African language technologies
    • Kathleen Siminyu, Jade Z. Abbott, Kólá Túbosún, Aremu Anuoluwapo, Blessing K. Sibanda, Kofi Yeboah, David Ifeoluwa Adelani, Masabata Mokgesi-Selinga, Frederick R. Apina, Angela Thandizwe Mthembu, Arshath Ramkilowan, Babatunde Oladimeji.
    • Patterns (Cell Press), Volume 4, Issue 1, January 2023. DOI: 10.1016/j.patter.2022.100654
  3. Corpus Development of Kiswahili Speech Recognition Test and Evaluation sets, Preemptively Mitigating Demographic Bias Through Collaboration with Linguists
    • Kathleen Siminyu, Kibibi Mohamed Amran, Abdulrahman Ndegwa Karatu, Mnata Resani, Mwimbi Makobo Junior, Rebecca Ryakitimbo, Britone Mwasaru.
    • Proceedings of the 5th Workshop on the Use of Computational Methods in the Study of Endangered Languages (ComputEL-5), alongside ACL 2022. ACL Anthology
  4. Phoneme Recognition through Fine Tuning of Phonetic Representations: a Case Study on Luhya Language Varieties
    • Kathleen Siminyu, Xinjian Li, Antonios Anastasopoulos, David Mortensen, Michael R. Marlo, Graham Neubig.
    • Interspeech 2021. arXiv:2104.01624
  5. AI4D - African Language Program
    • Kathleen Siminyu, Godson Kalipe, Davor Orlic, Jade Abbott, Vukosi Marivate, Sackey Freshia, Prateek Sibal, Bhanu Neupane, David I. Adelani, Amelia Taylor, Jamiil Toure ALI, Kevin Degila, Momboladji Balogoun, Thierno Ibrahima DIOP, Davis David, Chayma Fourati, Hatem Haddad, Malek Naski.
    • 2nd AfricaNLP Workshop alongside EACL 2021. arXiv:2104.02516
  6. AI4D - African Language Dataset Challenge
    • Kathleen Siminyu, Sackey Freshia, Jade Abbott, Vukosi Marivate.
    • Widening Natural Language Processing (WiNLP) Workshop alongside ACL 2020. arXiv:2007.11865

Co-Authored & Collaborative Works

  1. Towards Best Practices for Open Datasets for LLM Training
    • Stefan Baack, Stella Biderman, Kasia Odrozek, Aviya Skowron, Ayah Bdeir, Jillian Bommarito, Jennifer Ding, Maximilian Gahntz, Paul Keller, Pierre-Carl Langlais, Greg Lindahl, Sebastian Majstorovic, Nik Marda, Guilherme Penedo, Maarten Van Segbroeck, Jennifer Wang, Leandro von Werra, Mitchell Baker, Julie Belião, Kasia Chmielinski, Marzieh Fadaee, Lisa Gutermuth, Hynek Kydlíček, Greg Leppert, EM Lewis-Jong, Solana Larsen, Shayne Longpre, Angela Oduor Lungati, Cullen Miller, Victor Miller, Max Ryabinin, Kathleen Siminyu, Andrew Strait, Mark Surman, Anna Tumadóttir, Maurice Weber, Rebecca Weiss, Lee White, Thomas Wolf.
    • arXiv preprint, 2025. arXiv:2501.00000
  2. Road map for research on responsible artificial intelligence for development (AI4D) in African countries: The case study of agriculture
    • Arthur Gwagwa, Emre Kazim, Patti Kachidza, Airlie Hilliard, Kathleen Siminyu, Matthew Smith, John Shawe-Taylor.
    • Patterns (Cell Press), Volume 2, Issue 1, January 2021. DOI: 10.1016/j.patter.2020.100181
  3. Tusom2021: A Phonetically Transcribed Speech Dataset from an Endangered Language for Universal Phone Recognition Experiments
    • David Mortensen, Jordan Picone, Xinjian Li, Kathleen Siminyu.
    • Interspeech 2021. arXiv:2104.00824
  4. Participatory Research for Low-resourced Machine Translation: A Case Study in African Languages
    • Wilhelmina Nekoto, Vukosi Marivate, ..., Kathleen Siminyu, ..., Abdallah Bashir (Masakhane).
    • Findings of Empirical Methods in Natural Language Processing (EMNLP 2020). (Recognized with the Wikimedia Research Award of the Year 2021). arXiv:2010.02353
  5. Masakhane - Machine Translation For Africa
    • Iroro Orife, Julia Kreutzer, Blessing Sibanda, Daniel Whitenack, Kathleen Siminyu, Laura Martinus, Jamiil Toure Ali, Jade Abbott, Vukosi Marivate, et al.
    • 1st AfricaNLP Workshop alongside ICLR 2020.

Datasets & Open Source Contributions

  • Kiswahili Common Voice Dataset (Mozilla Foundation): Curated and validated over 1,000 hours of open Kiswahili voice recordings with acoustic evaluation models and an open API.
  • Bukusu and Saamia Phone Recognition Sets (CMU NeuLab): Transcribed acoustic phonetic datasets for two varieties of the Luhya cluster of western Kenya and eastern Uganda.
  • AI4D Language Challenges (Zindi & IDRC): Designed and coordinated crowdsourcing competitions resulting in over 15 open African language datasets.