gemmozero/ai-language-preservation-2026

1{"0": "google_endangered", "1": "Google Project Euphonia", "2": "Multiple", "3": "critically_endangered", "4": "Speech recognition for atypical speech", "5": "Google", "6": "Unknown", "7": "Open dataset for 1000+ languages", "8": "https://sites.research.google/euphonia/", "9": "2026-09-25"}2{"0": "meta_nllb", "1": "Meta NLLB", "2": "200 languages", "3": "low_resource", "4": "No Language Left Behind translation model", "5": "Meta", "6": "Unknown", "7": "Open model covering 200 languages", "8": "https://ai.meta.com/research/no-language-left-behind/", "9": "2026-09-25"}3{"0": "maori_ai", "1": "Te Hiku Media", "2": "Maori — 50K speakers", "3": "endangered", "4": "ASR + TTS for Maori language", "5": "NZ Govt + iwi", "6": 2000000.0, "7": "First indigenous-controlled AI language model", "8": "https://tehiku.nz/te-hiku-tech/", "9": "2026-09-25"}4{"0": "cherokee_tts", "1": "Cherokee Nation", "2": "Cherokee — 2000 speakers", "3": "critically_endangered", "4": "TTS to teach Cherokee pronunciation", "5": "Cherokee Nation", "6": 500000.0, "7": "First Cherokee TTS system", "8": "https://cherokee.org/", "9": "2026-09-25"}5{"0": "masakhane", "1": "Masakhane", "2": "African languages 500+", "3": "under_resourced", "4": "NLP benchmarks for African languages", "5": "Open collective", "6": 300000.0, "7": "50+ language NLP models", "8": "https://www.masakhane.io/", "9": "2026-09-25"}6{"0": "ai4bharat", "1": "AI4Bharat", "2": "22 Indian languages", "3": "low_resource", "4": "ASR+TTS for Indian languages", "5": "Indian govt+IIT Madras", "6": 5000000.0, "7": "Indic language models covering 1.4B speakers", "8": "https://ai4bharat.iitm.ac.in/", "9": "2026-09-25"}7{"0": "language_preservation_grants", "1": 450000000.0, "2": 12.3, "3": "UNESCO,Ford Foundation,Google", "4": 2024}8{"0": "indigenous_ai_tools", "1": 180000000.0, "2": 34.1, "3": "Te Hiku,Masakhane,AI4Bharat", "4": 2024}9{"0": "government_language_programs", "1": 2300000000.0, "2": 8.7, "3": "National govts worldwide", "4": 2024}10