Back to Home
NileAGI

Blog

Insights and updates from our research and development

Featured

Speaking Kiswahili Back

Paste Kiswahili, hear it spoken. A research preview of Swahili text-to-speech, with a try box and weights on Hugging Face.

•Zephania Reuben and Isack Odero
Read more

All Articles

Hearing Kiswahili

Record or upload clear Kiswahili and read the transcript back. A research preview of Swahili speech-to-text on Hugging Face.

•Zephania Reuben and Isack Odero

Giving Sukuma a Voice

Paste Sukuma, hear it come back. Quality and lite voices for everyday machines, with a live try-out.

•Zephania Reuben and Isack Odero

Listening for Sukuma

Turn Sukuma speech into Sukuma text. Quality and lite checkpoints you can try live, without routing the voice through another language first.

•Zephania Reuben and Isack Odero

All Languages Matter

Kamusi started with Swahili. This next step is small Swahili-to-Sukuma translators for everyday machines, with a live try-out and the technical report.

•Zephania Reuben and Isack Odero

NILEAGI-SUB: Measuring Swahili Understanding Where It Matters

We open SUB-MCQ EDUCATION: 2,569 Swahili school items, eight compact open models, and a sector-first evaluation programme featured in the Cohere Labs Open Science Community.

•Zephania Reuben and Isack Odero

Introducing LAI: Language Intelligence Platform for Culturally Aware AI

Announcing the Language Intelligence (LAI) Platform, a community-driven data collection and validation ecosystem designed to preserve languages and produce high-quality datasets for culturally aware AI systems.

•Zephania Reuben and Isack Odero

Breaking Free from Cloud Dependencies: The Delta CLI and Web UI

Exploring real-world use cases and the transformative power of offline AI with Delta CLI.

•Zephania Reuben and Isack Odero

NileAGI Builders' Space

Accelerating an Open AI Ecosystem. Join us in building the future of accessible machine intelligence.

•Zephania Reuben and Isack Odero

Introducing lm.c: A CPU Inference Engine for LLMs

A C99-based CPU inference engine enabling large language models to run with high efficiency and zero dependencies across all CPU architectures.

•Zephania Reuben and Isack Odero

Ellipse: An Efficient Deep Learning Framework for Low-Resource Environments

A minimalistic high-performance AI engine written in C, designed for accessibility and efficiency in resource-constrained environments.

•Zephania Reuben and Isack Odero

Delta: A Deep Dive into Accessible AI

Exploring how Delta CLI is transforming AI accessibility, particularly for regions with limited hardware resources.

•Zephania Reuben and Isack Odero