Real-Time Generative AI Experience for Qualcomm Snapdragon Summit
Challenge
Qualcomm needed an interactive demonstration that could showcase advanced AI capabilities running on Snapdragon hardware during the Snapdragon Summit 2023 event. The experience had to support real-time voice interaction, dynamic image generation, and responsive visual output while operating reliably in a live event environment.
The technical challenge involved integrating Whisper-based voice processing with generative AI workflows running directly on mobile hardware. The system also needed to coordinate communication between multiple components while maintaining fast response times and a smooth user experience during continuous public interaction.
Qualcomm needed an interactive demonstration that could showcase advanced AI capabilities running on Snapdragon hardware during the Snapdragon Summit 2023 event. The experience had to support real-time voice interaction, dynamic image generation, and responsive visual output while operating reliably in a live event environment.
The technical challenge involved integrating Whisper-based voice processing with generative AI workflows running directly on mobile hardware. The system also needed to coordinate communication between multiple components while maintaining fast response times and a smooth user experience during continuous public interaction.
Solution
A Mad Libs-style interactive AI application was designed and developed to combine voice interaction with real-time image generation. Users speak partial prompts through Whisper-powered voice recognition, and the system dynamically completes and processes those prompts using generative AI workflows.
Fast Stable Diffusion was integrated on Qualcomm’s Snapdragon 8 Gen 3 Mobile Platform to generate images directly on-device with render times under four seconds. A backend coordination layer connected the phone, voice-processing pipeline, and image-generation services to ensure synchronized execution across all components.
The application also included dynamic visual display support for external screens, optimized animations, and performance tuning for live event deployment.
A Mad Libs-style interactive AI application was designed and developed to combine voice interaction with real-time image generation. Users speak partial prompts through Whisper-powered voice recognition, and the system dynamically completes and processes those prompts using generative AI workflows.
Fast Stable Diffusion was integrated on Qualcomm’s Snapdragon 8 Gen 3 Mobile Platform to generate images directly on-device with render times under four seconds. A backend coordination layer connected the phone, voice-processing pipeline, and image-generation services to ensure synchronized execution across all components.
The application also included dynamic visual display support for external screens, optimized animations, and performance tuning for live event deployment.
Results
The system was successfully deployed at the Snapdragon Summit, where it operated as a live AI experience demonstrating Qualcomm’s on-device AI capabilities.
Users interacted with the platform through real-time voice commands while generated visuals appeared almost instantly on connected displays. The experience highlighted the performance of Fast Stable Diffusion and Whisper integration on Snapdragon hardware and contributed to strong attendee engagement during the event.
The project demonstrated reliable execution of real-time generative AI workflows in a production event setting while combining voice interaction, AI image generation, and mobile hardware acceleration into a single system.
The system was successfully deployed at the Snapdragon Summit, where it operated as a live AI experience demonstrating Qualcomm’s on-device AI capabilities.
Users interacted with the platform through real-time voice commands while generated visuals appeared almost instantly on connected displays. The experience highlighted the performance of Fast Stable Diffusion and Whisper integration on Snapdragon hardware and contributed to strong attendee engagement during the event.
The project demonstrated reliable execution of real-time generative AI workflows in a production event setting while combining voice interaction, AI image generation, and mobile hardware acceleration into a single system.
