File size: 3,003 Bytes
48004e5 e5dbee8 48004e5 e5dbee8 48004e5 e5dbee8 13ec987 e5dbee8 94142e2 772fa0c e5dbee8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 | ---
title: lab.et — The AI Lab of Ethiopia
emoji: 🎙️
colorFrom: red
colorTo: yellow
pinned: true
license: cc-by-4.0
---
# 🇪🇹 lab.et — The AI Lab of Ethiopia
<div align="center">
**We build artificial intelligence for Ethiopian languages.**
*በእኛ ቋንቋ የሚሰራ አስተውሎት።*
[🌐 lab.et](https://lab.et) · [🎙️ voice.et](https://voice.et) · [✨ yakal.et](https://yakal.et)
</div>
---
## Who we are
[lab.et](https://lab.et) is an artificial intelligence lab based in **Addis Ababa, Ethiopia**. Amharic is spoken by over 50 million people — yet it remains one of the world's most under-resourced languages in AI. We exist to change that.
We collect our own datasets. We train our own models. We run our own infrastructure. No rented intelligence, no foreign data flows — Ethiopian AI, built in Ethiopia.
## What we build
### 🎙️ [voice.et](https://voice.et) — Speech AI
Automatic speech recognition and natural text-to-speech for **Amharic, Afaan Oromoo, Tigrinya, Sidamo and Wolaytta** — in production, serving real users through APIs and live voice products.
### ✨ [yakal.et](https://yakal.et) — The AI Console
Chat, images, video, voice and an AI site builder — in Amharic, Afaan Oromoo and English, **prepaid in Birr** via Telebirr. Built for Ethiopian creators, developers and businesses.
### 🏢 Enterprise & Research
Custom speech models tuned to your domain, on-premise and sovereign deployments that keep Ethiopian data inside Ethiopia, and ethically-sourced language corpora for research institutions.
## 📦 Open releases
| Release | What it is |
|---|---|
| [azemari](https://huggingface.co/Lab-et/azemari) | 🎙️ **Our flagship TTS** — multilingual Ethiopian voice: Amharic, Afaan Oromoo, Tigrinya · emotion control · voice cloning · complete checkpoint, ready to run |
| [omnivoice-amharic](https://huggingface.co/Lab-et/omnivoice-amharic) | Our in-house Amharic text-to-speech system |
| [amharic-corpus](https://huggingface.co/datasets/Lab-et/amharic-corpus) | **408.5 hours** of clean synthetic Amharic speech (146,359 clips) — generated entirely by OmniVoice |
| [azemari-samples](https://huggingface.co/datasets/Lab-et/azemari-samples) | Listening samples for Azemari — languages, emotions, voice cloning |
| *more coming soon* | datasets, models, benchmarks for Ethiopian languages |
## By the numbers
- 🗣️ **5 languages** supported in production ASR
- 📼 **~193k** speech utterances in our training corpora
- 🔬 **In-house** model training, evaluation and research pipelines
- 🏭 **Production-proven** — live products serving real users since 2025
## Work with us
Building for the Ethiopian market? Banks, telcos, government services, NGOs — reach us:
- 🌐 [lab.et](https://lab.et)
- ✉️ [admin@lab.et](mailto:admin@lab.et)
We read everything.
---
<div align="center">
**lab.et** · Built in Addis Ababa · *በአዲስ አበባ የተሰራ* 🇪🇹
</div>
|