--- title: a emoji: ๐Ÿš€ short_description: Open AI models and datasets for the Uzbek language. --- # Open Limi We build and improve AI models for the Uzbek language, so that Uzbek speakers get the same quality of AI as speakers of any other major language. ## What we do - ๐ŸŽ™๏ธ **Speech Recognition (ASR)** โ€“ accurate transcription for Uzbek - ๐Ÿ”Š **Text-to-Speech (TTS)** โ€“ natural, expressive Uzbek voices - ๐ŸŒ **Translation & Speech-to-Speech** โ€“ Uzbek โ†” English/Russian and more - ๐Ÿ’ฌ **Language Models** โ€“ fine-tuned LLMs and embeddings for Uzbek - ๐Ÿ“Š **Datasets & Benchmarks** โ€“ clean, documented data and evaluation sets ## Our approach 1. Train new models where none exist. 2. Fine-tune and improve existing open models for Uzbek. 3. Release with clear model cards, benchmarks, and licenses. ## Contact ๐ŸŒ ยท ๐Ÿ“ง openlimiuz@gmail.com ยท ๐Ÿ’ผ