Tencent Gander: AI คุยเสียงพร้อมทำงานเบื้องหลัง แบบแยกสมอง

Tencent unveiled project Gander in early November, showcasing an AI that can converse in real-time, understand speech, recognize images, and work in the background simultaneously without interrupting the conversation. This innovative approach is based on a dual-brain system inspired by the anatomy of the human cerebellum and brain, designed to handle speech interactions and background tasks separately.

The cerebellum, or "small brain," manages real-time voice conversations, visual input from sent photos, and rapid response generation. In contrast, the larger brain focuses on more demanding tasks like data retrieval, coding, or invoking external tools. These two components are connected via a queue manager, allowing the small brain to continue interacting with the user while the large brain completes heavy tasks.

Currently, Tencent offers no public online service for Gander; users must download and set up the system independently, requiring three graphics cards and significant memory. Despite the high memory requirements, the project remains research-focused rather than commercial, with models available for download under the permissive Apache 2.0 license.

While the system demonstrates promising voice interaction capabilities, it still lags behind commercial AI services like GPT-Realtime in performance. The main advantage lies in the open-source nature, enabling developers to customize and run their own instances at a lower cost compared to proprietary solutions. However, the current performance and setup complexity make it more suitable for researchers and developers interested in exploring the mechanics of dual-brain AI rather than everyday users seeking a ready-to-use voice assistant.