Developing story · 2 episodes · 21 Oct to 26 Oct
Open Source AI Multilingual Model Race
The saga tracks the rapid advancement of open-source AI models, beginning with IBM's multilingual speech recognition tool and moving to Tencent's large-scale multilingual translation system. The latest episode highlights the release of a 30-billion parameter model designed for efficient and accurate cross-lingual communication.
-
Tencent/Hy-SimulMT-30B-A3B · Hugging Face
The model processes streaming audio and outputs both translated text and speech while cloning the speaker's voice in real time. Tencent reports a StreamLAAL of approximately 3 seconds on the RealSI…
1 source primary source -
IBM releases Granite Speech 5.0 2B for multilingual speech recognition
IBM announced the Granite Speech 5.0 2B model, a 2 billion‑parameter speech‑language system aimed at automatic speech recognition (ASR) and automatic speech translation (AST) in English, French,…
2 sources primary source
Who and what
IBMHugging FaceTencentTencent HunyuanModelScopeGranite Speech 5.0 2Bgranite-4.0-1b-baseGranite GuardianHy-SimulMT-30B-A3BHy-SimulMT-30B-A3B-InterleaveMao ZhengBo Lv