VoiceLab

VoiceLab works on Uzbek speech AI.

We build and maintain models, datasets, and tools for Uzbek voice technology — from speech recognition to text-to-speech.

What we are building

• Lison — Uzbek text-to-speech and voice generation.
• Sado — Uzbek speech-to-text and speech understanding.

What you can find here

• TTS models and voice-cloning experiments.
• STT models and alignment models.
• Curated Uzbek speech datasets.
• Dataset cards and model cards with clear provenance notes.

Learn more: voicelab.uz