badrex/w2v-bert-2.0-kikuyu-asr

Automatic Speech Recognition for Kikuyu

Model Description 🍍

This model is a fine-tuned version of Wav2Vec2-BERT 2.0 for Kikuyu automatic speech recognition (ASR). It was trained on the 100+ hours of transcribed speech, covering Health, Government, Finance, Education, and Agriculture domains. The in-domain WER for this ASR model is below 25.0%.

• Developed by: Badr al-Absi
• Model type: Speech Recognition (ASR)
• Language: Kikuyu (kik)
• License: CC-BY-4.0
• Finetuned from: facebook/w2v-bert-2.0

Model Architecture

• Base model: Wav2Vec2-BERT 2.0
• Architecture: transformer-based with convolutional feature extractor
• Parameters: ~600M (inherited from base model)
• Objective: connectionist temporal classification (CTC)

Funding

The development of this model was supported by CLEAR Global and Gates Foundation.

Citation

@misc{w2v_bert_kikuyu_asr,
  author = {Badr M. Abdullah},
  title = {Adapting Wav2Vec2-BERT 2.0 for Kikuyu ASR},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/badrex/w2v-bert-2.0-swahili-asr}
}

Model Card Contact

For questions or issues, please contact via the Hugging Face model repository in the community discussion section.