π About Us
Aaivu β from the Tamil word for research β is an open-source organization building AI that enhances and enriches the lives of the communities who benefit from it. We focus on low-resource languages and under-served problem domains, releasing datasets, models, and tools openly so researchers everywhere can build on our work.
π¬ What We Work On
- π£οΈ Speech & Emotion Recognition β datasets and models for under-represented languages
- π Low-Resource NLP & Machine Translation β Tamil, Sinhala, and beyond
- π¬ Dialogue Systems β policy optimization in low-resource settings
- π Time-Series Forecasting β transformer architectures for panel data
- π§ Explainable & Scalable AI β interpretability that scales
β Featured Work
| Project | What it is |
|---|---|
| EmoTa | Open-access Tamil Speech Emotion Recognition dataset β 936 utterances, 22 native speakers, 5 emotions |
| KuralHub | Multilingual collection of Speech Emotion Recognition datasets |
| Panelformer | Transformer for accurate, scalable panel time-series forecasting |
| EngβSin NMT | English β Sinhala Neural Machine Translation |
π€ Get Involved
We love open source. Explore our datasets and models right here on Hugging Face, star our repositories on GitHub, or say hello at hello@aaivu.org.
π Building AI for everyone β in every language.