ivrit.ai: Building World-Class AI for Hebrew

Our 41th DataTalks meetup will be hosted by AI21, and will focus on how ivrit.ai, a non-profit, unfunded, volunteer-based initiative, built the world's best Hebrew speech-to-text engine.

We will discuss the challenges, datasets, models, and next steps to make Hebrew a first-class AI citizen.

🔴 Location: AI21 offices, Leonardo da Vinci St 14, 4th floor, Tel Aviv-Yafo.

🔴 Agenda:
18:00-18:30 – Gathering, Mingling, etc. (30 minutes)
18:30-18:50 – Opening Session: AI21
18:50-19:10 – The Story of ivrit.ai So Far (ivrit.ai team: Yair Lifshitz, Yanir Marmor & Dr. Kinneret Misgav)

Technological Insights
19:10-19:30 – A Language Modeling Approach to Diacritic-Free Hebrew TTS (Arnon Turetzky, CS PhD student at HUJI)
19:30-19:45 – Break & Refreshments (15 minutes)

AI and Legal – Perspectives (30 minutes)
19:45-20:00 – Legalese and Linguistics in the ivrit.ai License (Adv. Eli Greenbaum, Arnon Tadmor-Levy)
20:00-20:15 – Legal Aspects of AI Training (Dr. Lital Helman, Law and Technology Expert)

Artistic Segment (10 minutes)
20:15-20:25 – Spoken Word: My Family and Other Startups (Efrat Aran, Product Data Scientist, AI21)

Linguistic Perspectives (20 minutes)
20:25-20:45 – A Linguistic Perspective on Spoken Language (Nadav Matalon, Interactional Linguist, HUJI and Weizmann Institute of Science)

Closing Session (40 minutes)
20:45-21:25 – Announcement of New Developments! and Closing: Where Do We Want to Go From Here? (ivrit.ai team: Yair Lifshitz, Yanir Marmor & Dr. Kinneret Misgav)