KRAFTON has released A.X K2 Raon-Speech, a bilingual English/Korean speech language model with 21.2 billion total parameters and 3.5 billion active parameters. This multimodal model integrates a text backbone from SK Telecom's A.X K2 Light 20B-A3B, along with independently trained speech encoder and neural audio codec modules developed by KRAFTON AI. The model demonstrates strong performance in various speech tasks, including speech recognition, synthesis, and question answering, achieving top rankings among publicly available models under 30 billion parameters for both Korean and English. AI
IMPACT This model's bilingual capabilities and multi-task performance could advance speech technology integration in diverse applications.
RANK_REASON Release of a new speech language model with detailed technical specifications and benchmark results. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
- API-Bank
- A.X K2 Light 20B-A3B
- A.X K2 Raon-Speech
- FunctionChat-Bench
- Google Colab
- Kaggle
- Krafton
- KRAFTON/A.X-K2-Raon-Speech-21B-A3B
- KsponSpeech
- KVoiceBench
- LibriSpeech
- Mimi
- Raon-Speech-9B
- SK Telecom
- SpeechBrain
- VoiceBench
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
