Elamur Research⌕
☰
home / papers / 2026 / Voice Activation Detection for Transcription of Indigenous Languages

Voice Activation Detection for Transcription of Indigenous Languages

Summary

Automatic transcription pipelines for endangered languages usually begin by separating speech from silence and background sound. However, off-the-shelf voice activity detection (VAD) tools were developed mostly on urban or studio-like audio and had not been tested on fieldwork recordings. The authors evaluate four VAD systems on recordings from two unrelated Indigenous languages: Bribri (Chibchan, Costa Rica) and Cook Islands Māori (Polynesian). The systems are PyDub (amplitude thresholding), WebRTC VAD (GMM-based), Silero VAD and SpeechBrain VAD (both neural). Their outputs were compared with manually verified ELAN segmentations, scored in 10 ms frames using accuracy, precision, recall and F1. SpeechBrain, which combines a CRDNN with energy-based post-processing, clearly performed best, while purely energy-based detection was close to useless. All systems handled non-speech poorly and often treated rain, animals or appliance hum as voice. The study shows that the choice of VAD strongly affects documentation workflows. It also suggests that fieldwork soundscapes are underrepresented in current training data, which motivates fine-tuning on such audio.

Key points

  • Data: three single-speaker fieldwork recordings per language, about 17 minutes for Bribri and 73 minutes for Cook Islands Māori, with human-annotated ELAN files as the gold standard.
  • Frame-level accuracy: SpeechBrain VAD reached 89% (Bribri) and 94% (CIM). WebRTC VAD scored 84% and 73%, Silero VAD 55% and 74%, and PyDub only about 9–13%.
  • Speech F1 for SpeechBrain was 94% (Bribri) and 97% (CIM), but its non-speech F1 was just 29% and 46%. Every tool tended to over-label noise as speech.
  • Silero performed much worse on the noisier Bribri audio. The authors attribute this to its training on urban, media-style speech, and they suggest WebRTC's reliance on low-level features makes it more robust there.
  • Planned follow-up includes testing pyannote.audio, Cobra VAD and WhisperX, fine-tuning on fieldwork audio, and adding more languages and lab-recorded controls, since the current sample is small.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.computel-1.19

@inproceedings{cotosolano2026voice,
  title = {{Voice Activation Detection for Transcription of Indigenous Languages}},
  author = {Coto-Solano, Rolando and Browning, Mikaela and Corrado, Thomas and Nicholas, Sally Akevai},
  booktitle = {Proceedings of the Ninth Workshop on the Use of Computational Methods in the Study of Endangered Languages (ComputEL-9)},
  year = {2026},
  doi = {10.18653/v1/2026.computel-1.19},
  url = {https://doi.org/10.18653/v1/2026.computel-1.19}
}

References

51 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Carter Anderson, Mien Nguyen, and Rolando Coto-Solano. 2025. Unsupervised, semi-supervised and llm-based morphological segmentation for bribri. In Proceedings of the Fifth Workshop on NLP for Indigenous Languages of the Americas (AmericasNLP), pages 63–76. find
  2. Max Bain, Jaesung Huh, Tengda Han, and Andrew Zisserman. 2023. WhisperX: Time-Accurate Speech Transcription of Long-Form Audio. In Proc. find
  3. Hervé Bredin and Antoine Laurent. 2021. End-to-end speaker segmentation for overlap-aware resegmentation. In Proc. find
  4. Luis Chiruzzo, Pavel Denisov, Alejandro Molina-Villegas, Silvia Fernandez-Sabido, Rolando Coto-Solano, Marvin Agüero-Torales, Aldo Alvarez, Samuel Canul-Yah, Lorena Hau-Ucán, Abteen Ebrahimi, Robert Pugh, Arturo Oncevay, Shruti Rijhwani, Katharina von der Wense, and Manuel Mager. 2024. Findings of t… find
  5. Rolando Coto-Solano. 2021. Explicit tone transcription improves ASR performance in extremely low-resource languages: A case study in Bribri. In Proceedings of the First Workshop on Natural Language Processing for Indigenous Languages of the Americas, pages 173–184, Online. Association for Computatio… find
  6. Rolando Coto-Solano. 2022. Evaluating word embeddings in extremely under-resourced languages: A case study in Bribri. In Proceedings of the 29th International Conference on Computational Linguistics, pages 4455–4467. find
  7. Rolando Coto-Solano and Sofía Flores-Solórzano. 2016. Alineación forzada sin entrenamiento para la anotación automática de corpus orales de las lenguas indígenas de costa rica. Kánina, 40(4):175–199. find
  8. Rolando Coto-Solano, Tai Wan Kim, Alexander Jones, and Sharid Loáiciga. 2024. Multilingual Models for ASR in Chibchan Languages. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers),… find
  9. Rolando Coto-Solano, Daisy Li, Manoela Teleginski Ferraz, Olivia Sasse, Cha Krupka, Sharid Loáiciga, and Sally Akevai Tenamu Nicholas. 2025. Diacritic restoration for low-resource indigenous languages: Case study with bribri and cook islands māori. arXiv preprint arXiv:2512.19630. find our card
  10. Rolando Coto-Solano, Sharid Loáiciga, and Sofía Flores-Solórzano. 2021. Towards universal dependencies for bribri. In Proceedings of the Fifth Workshop on Universal Dependencies (UDW, SyntaxFest 2021), pages 16–29. find
  11. Rolando Coto-Solano, Sally Akevai Nicholas, Samiha Datta, Victoria Quint, Piripi Wills, Emma Ngakuravaru Powell, Liam Koka’ua, Syed Tanveer, and Isaac Feldman. 2022a. Development of automatic speech recognition for the documentation of Cook Islands Māori. In Proceedings of the Thirteenth Language R… find
  12. Rolando Coto-Solano, Sally Akevai Nicholas, Brittany Hoback, and Gregorio Tiburcio Cano. 2022b. Managing data workflows for untrained forced alignment: examples from Costa Rica, Mexico, the Cook Islands, and Vanuatu. The Open Handbook of Linguistic Data Management, 35. find
  13. Rolando Coto-Solano, Sally Akevai Nicholas, and Samantha Wray. 2018. Development of natural language processing tools for Cook Islands Māori. In Proceedings of the Australasian Language Technology Association Workshop 2018, pages 26–33. find
  14. Ona De Gibert, Robert Pugh, Ali Marashian, Raul Vazquez, Abteen Ebrahimi, Pavel Denisov, Enora Rice, Edward Gow-Smith, Juan Prieto, Melissa Robles, Rubén Manrique, Oscar Moreno, Angel Lino, Rolando Coto-Solano, Aldo Alvarez, Marvin Agüero-Torales, John E. Ortega, Luis Chiruzzo, Arturo Oncevay, and 3… find
  15. G. Durantin, B. Foley, N. Evans, and J. Wiles. 2017. Transcription survey. Paper presented at the Australian Linguistic Society Annual Conference. find
  16. Abteen Ebrahimi, Manuel Mager, Arturo Oncevay, Vishrav Chaudhary, Luis Chiruzzo, Angela Fan, John Ortega, Ricardo Ramos, Annette Rios, Ivan Vladimir Meza Ruiz, Gustavo Giménez-Lugo, Elisabeth Mager, Graham Neubig, Alexis Palmer, Rolando Coto-Solano, Thang Vu, and Katharina Kann. 2022a. AmericasNLI:… find our card
  17. Abteen Ebrahimi, Manuel Mager, Shruti Rijhwani, Enora Rice, Arturo Oncevay, Claudia Baltazar, María Cortés, Cynthia Montaño, John E. Ortega, Rolando Coto-solano, Hilaria Cruz, Alexis Palmer, and Katharina Kann. 2023a. Findings of the AmericasNLP 2023 shared task on machine translation into indigenou… find
  18. Abteen Ebrahimi, Manuel Mager, Adam Wiemerslage, Pavel Denisov, Arturo Oncevay, Danni Liu, Sai Koneru, Enes Yavuz Ugan, Zhaolin Li, Jan Niehues, Monica Romero, Ivan G Torre, Tanel Alumäe, Jiaming Kong, Sergey Polezhaev, Yury Belousov, Wei-Rui Chen, Peter Sullivan, Ife Adebara, and 15 others. 2022b.… find
  19. Abteen Ebrahimi, Arya D. McCarthy, Arturo Oncevay, John E. Ortega, Luis Chiruzzo, Gustavo Giménez-Lugo, Rolando Coto-Solano, and Katharina Kann. 2023b. Meeting the needs of low-resource languages: The value of automatic alignments via pretrained models. In Proceedings of the 17th Conference of the E… find
  20. Isaac Feldman and Rolando Coto-Solano. 2020. Neural machine translation models with back-translation for the extremely low-resource indigenous language Bribri. In Proceedings of the 28th International Conference on Computational Linguistics, pages 3965– 3976, Barcelona, Spain (Online). International… find
  21. Sofía Flores-Solórzano. 2010. Teclado Chibcha: Un software lingüístico para los sistemas de escritura de las lenguas bribri y cabécar. Revista de Filología y Lingüística de la Universidad de Costa Rica, 36(2):155–161. find
  22. Sofía Flores-Solórzano. 2017. Corpus oral pandialectal de la lengua bribri. http://bribri.net. ↗
  23. Sofía Flores-Solórzano. 2017. Un primer corpus pandialectal oral de la lengua bribri y su anotacion morfologica con base en el modelo de estados finitos. Ph.D. thesis, Universidad Autónoma de Madrid. find
  24. Sofía Flores-Solórzano. 2019. La modelización de la morfología verbal bribri. Procesamiento del Lenguaje Natural, 62:85–92. find
  25. Sofía Flores-Solórzano and Rolando Coto-Solano. 2017. Comparison of two forced alignments systems for aligning bribri speech. CLEI Electronic Journal, 20(1):2–1. find
  26. Ben Foley, Joshua T Arnold, Rolando Coto-Solano, Gautier Durantin, T Mark Ellison, Daan Van Esch, Scott Heath, Frantisek Kratochvil, Zara Maxwell-Smith, David Nash, and 1 others. 2018. Building Speech Recognition Systems for Language Documentation: The CoEDL Endangered Language Pipeline and Inferenc… find
  27. INEC. 2011. X Censo Nacional de Población y VI de Vivienda 2011 - Territorios Indígenas - Principales Indicadores Demográficos y Socioeconómicos. find
  28. Jesin James, Rolando Coto-Solano, Sally Akevai Nicholas, Joshua Zhu, Bovey Yu, Fuki Babasaki, Jenny Tyler Wang, and Nicholas Derby. 2024. Development of community-oriented text-to-speech models for māori ‘avaiki nui (cook islands māori). In Proceedings of the 2024 Joint International Conference on… find
  29. Alex Jones, Rolando Coto-Solano, and Guillermo González Campos. 2023. TalaMT: Multilingual machine translation for Cabécar-Bribri-Spanish. In Proceedings of the 3rd Workshop on Multi-lingual Representation Learning (MRL), pages 106–117, Singapore. Association for Computational Linguistics. find
  30. Katharina Kann, Abteen Ebrahimi, Manuel Mager, Arturo Oncevay, John E. Ortega, Annette Rios, Angela Fan, Ximena Gutierrez-Vasques, Luis Chiruzzo, Gustavo A. Giménez-Lugo, Ricardo Ramos, Ivan Vladimir Meza Ruiz, Elisabeth Mager, Vishrav Chaudhary, Graham Neubig, Alexis Palmer, Rolando Coto-Solano, an… find
  31. Sarah Karnes, Rolando Coto-Solano, and Sally Akevai Nicholas. 2023. Towards universal dependencies in cook islands māori. In Proceedings of the Sixth Workshop on the Use of Computational Methods in the Study of Endangered Languages, pages 124–129. find
  32. Jessica Karson and Rolando Coto-Solano. 2024. Morphological Tagging in Bribri Using Universal Dependency Features. In Proceedings of the 4th Workshop on Natural Language Processing for Indigenous Languages of the Americas (AmericasNLP 2024), pages 56–66. find
  33. Haakon Krohn. 2020. Elaboración de una base de datos en XML para un diccionario bribri–español español– bribri en la web. Porto das Letras, 6(3):38–58. find
  34. Haakon S. Krohn. 2021. Diccionario digital bilingüe bribri. http://www.haakonkrohn.com/bribri. ↗
  35. Manuel Mager, Arturo Oncevay, Abteen Ebrahimi, John Ortega, Annette Rios, Angela Fan, Ximena Gutierrez-Vasques, Luis Chiruzzo, Gustavo Giménez-Lugo, Ricardo Ramos, Ivan Vladimir Meza Ruiz, Rolando Coto-Solano, Alexis Palmer, Elisabeth Mager-Hois, Vishrav Chaudhary, Graham Neubig, Ngoc Thang Vu, and… find our card
  36. Ministry of Finance and Economic Management, Government of the Cook Islands. 2021. find
  37. Census 2021: Key findings. https://www.mfem.gov.ck/statistics/census-and-surveys/census/ 267-census-2021. ↗
  38. Sally Akevai Nicholas. 2012. Te Vairanga Tuatua o te Te Reo Māori o te Pae Tonga: Cook Islands Māori (Southern dialects). find
  39. Sally Akevai Nicholas and Rolando Coto-Solano. 2019. Glottal variation, teacher training and language revitalization in the cook islands. In Proceedings of the 19th International Congress of Phonetic Sciences, University of Melbourne, Australia, pages 3602–3606. find
  40. Sally Akevai Te Namu Nicholas. 2018. Language contexts: Te Reo Māori o te Pae Tonga o te Kuki Airani also known as Southern Cook Islands Māori. Language Documentation and Description, 15:64. find
  41. Picovoice. 2024. Cobra: On-device voice activity detection engine. https://github.com/Picovoice/ cobra. ↗
  42. Emily Prud’hommeaux, Robbie Jimerson, Richard Hatcher, and Karin Michelson. 2021. Automatic Speech Recognition for Supporting Endangered Language Documentation. Language Documentation & Conservation, 15. find
  43. Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2023. Robust speech recognition via large-scale weak supervision. Proceedings of the 40th International Conference on Machine Learning (ICML), 202:28492– 28518. find our card
  44. Mirco Ravanelli, Titouan Parcollet, Peter Plantinga, Aku Rouhe, Samuele Cornell, Loren Lugosch, Cem Subakan, Nauman Dawalatabad, Abdelwahab Heba, Jianyuan Zhong, Ju-Chieh Chou, Sung-Lin Yeh, Szu-Wei Fu, Chien-Feng Liao, Elena Rastorgueva, François Grondin, William Aris, Hwidong Na, Yan Gao, and 2 ot… — [arXiv:2106.04624](https://arxiv.org/abs/2106.04624) find
  45. James Robert. 2011. Pydub. https://github.com/ jiaaro/pydub. ↗
  46. Carlos Sánchez Avendaño. 2013. Lenguas en peligro en Costa Rica: vitalidad, documentación y descripción. Revista Káñina, 37(1):219–250. find
  47. J. Shi, J. D. Amith, R. Castillo García, E. G. Sierra, K. Duh, and S. Watanabe. 2021. Leveraging end-to-end ASR for endangered language documentation: An empirical study on Yolóxochitl Mixtec. Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics… find
  48. Silero Team. 2021. Silero vad: pre-trained enterprise-grade voice activity detector (vad), number detector and language classifier. https://github.com/ snakers4/silero-vad. ↗
  49. Sofía Flores Solórzano. 2009. Los mamiferos en la clasi-ficacion etnobiologica de la comunidad de amubre. Estudios de Lingüística Chibcha, 28:7–47. find
  50. John Wiseman. 2016. py-webrtcvad: Python interface to the webrtc voice activity detector. https://github.com/wiseman/py-webrtcvad. ↗
  51. Peter Wittenburg, Hennie Brugman, Albert Russel, Alex Klassmann, and Han Sloetjes. 2006. ELAN: a professional framework for multimodality research. In Proceedings of the Fifth International Conference on Language Resources and Evaluation (LREC’06), Genoa, Italy. European Language Resources Associati… find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/solano-2026-voice-activation-detection-transcription-indigenous/