Insta360 GO Ultra Integrates AI Voice Assistant Powered by Google's Gemini
Insta360 announced on August 7th that it is launching a powerful AI Voice Assistant for its GO Ultra thumb camera. This new feature aims to transform the device from a simple capture tool into an intelligent companion for users on the go.
A key aspect of this implementation is its region-specific approach to leveraging large language models (LLMs). Alibaba's Qwen LLM will power the AI assistant for users in mainland China, while those in Hong Kong, Macau, Taiwan, and other international markets will use Google's Gemini. The results, such as translations, can be broadcast directly through the camera's built-in speaker.

The system is built upon Insta360's proprietary R&D, which integrates the multimodal capabilities of Alibaba's Qwen model and its photo Q&A functions. The architecture employs a hybrid model, where on-device processing handles tasks like voiceprint recognition and intent detection, while more complex operations such as Q&A, mode switching, and translation are managed in the cloud.
The AI assistant offers three primary functions, which can be activated with the voice command "Hey Kira" or by long-pressing the shutter button while the camera is off. These features include: Face-to-face Translation for real-time, two-way conversations in scenarios like asking for directions or ordering food; Visual Q&A, which uses the camera to identify and provide information about objects like signs, buildings, or dishes after receiving user permission; and Voice Q&A for general inquiries like weather or directions, which operates without activating the camera.

