智谱推出AI输入法

Core Insights - The article discusses the official release and open-sourcing of the GLM-ASR series voice recognition models by Zhipu, along with the launch of the Zhipu AI input method based on these models [1] Group 1: Product Launch - Zhipu has launched the cloud-based voice recognition model GLM-ASR-2512, which supports real-time voice-to-text conversion [1] - The company introduced the open-source SOTA edge-side voice model GLM-ASR-Nano-2512, which has a parameter count of 1.5 billion [1] - The Zhipu AI input method is designed to integrate the capabilities of the GLM-ASR series models, allowing users to interact via voice on desktop computers [1] Group 2: Features and Capabilities - The GLM-ASR-2512 model enables accurate voice-to-text transcription and can directly utilize large model capabilities within the input method for tasks such as translation, rewriting, and emotion conversion [1]