Imagine wandering through a city with your very own AI travel guide, ready to reveal the stories behind every landmark, monument, or hidden gem you encounter. With Google’s Talking Tours, this vision takes a significant step closer to becoming a reality.
Google Talking Tours offers a fascinating glimpse into the future of AI-driven travel guidance. Developed as part of a collaboration between Google Arts & Culture Lab and artist-in-residence Gaël Hugo, this experiment leverages generative audio and Google’s cutting-edge Gemini AI to provide dynamic, location-specific insights about cultural landmarks captured in Street View.
Talking Tours currently covers 55 major landmarks around the world, offering an AI-generated audio guide that provides insights based on the visual content of Street View panoramas. Users can explore a 360-degree view of a site, take a snapshot, and receive detailed commentary from the AI. Additionally, users can click an “ask a question” button to generate three contextual questions about the location, enhancing interactivity and personalized learning.
The technology relies on a blend of visual analysis and geospatial data. Gemini, Google’s multimodal large language model, processes the scene, combines it with GPS data, and crafts a descriptive script. This script is then converted into audio using a Google AI audio model, creating an immersive learning experience.
You can even explore individual museums and galleries, take snapshots of specific works of art or artifacts, and listen to Talking Tours' AI-generated narrated guides. Unfortunately, at this stage, the AI does not yet possess extensive knowledge of all exhibits in every museum and gallery worldwide. As a result, you may often receive a generic-sounding response rather than detailed information about a specific work of art or artifact. However, if the exhibit is particularly notable - for example, the Rosetta Stone in the British Museum - Talking Tours is able to generate a reasonably informed narration that describes and explains the selected exhibit.
No comments:
Post a Comment