Key Highlights
- Six hyper‑realistic metahuman avatars form a virtual band led by AR Rahman.
- Google Cloud supplies the AI stack – Veo 3, Imagen, Gemini 2.5 Pro – for real‑time animation and interaction.
- Each avatar embodies a distinct culture and musical style, backed by live human collaborators.
- The platform enables live fan dialogue, storytelling and adaptive performances.
- Secret Mountain showcases scalable, secure AI governance for future entertainment projects.
Detailed Insights
AR Rahman’s vision for Secret Mountain is to merge musical artistry with immersive technology, creating a metahuman band that feels like a real ensemble. Six avatars – Cara, Zen Tam, Blessing, and three others – are generated as 3‑D digital characters that mimic the gestures, facial expressions and vocal nuances of authentic musicians. Each avatar is linked to a team of singers, dancers and producers who provide the underlying audio and choreography, ensuring that the virtual performance stays true to Rahman’s compositional intent.
Google Cloud’s contribution centers on its AI infrastructure. Veo 3 drives live video synthesis, enabling the avatars to move fluidly in virtual studios. The Imagen model, together with Gemini Flash 2.5 (Nano Banana), delivers photorealistic textures and visual effects. Gemini 2.5 Pro serves as the conversational core, allowing avatars to answer fan questions, adapt to live inputs and maintain contextual continuity across interactions. The integration of these models allows each avatar to sing, improvise, and respond in real time, blurring the line between virtual and human performance.
Beyond music, Secret Mountain is engineered as an interactive narrative platform. Audiences can trigger songs, ask for backstory, and even influence the development of the storyline. This level of engagement is made possible by a scalable, cloud‑native architecture that also incorporates content‑governance tools, ensuring that all interactions remain safe and aligned with copyright and brand guidelines. The initiative exemplifies how large‑scale AI can be deployed responsibly in consumer entertainment.
Key Concepts
- Metahuman – a highly detailed, AI‑generated digital human capable of realistic motion and expression.
- Veo 3 – a Google Cloud AI service that produces real‑time video rendering and avatar animation.
- Gemini 2.5 Pro – a multimodal AI model that powers natural‑language dialogue and contextual understanding.
- Digital Avatars – virtual characters that interact with users and perform tasks or entertainment.
- Content Governance – systematic controls that manage the quality, safety and compliance of user‑generated or AI‑driven content.