Building Your First Multimodal Gen AI App 🚀
summary
A hands‑on app is built live: a Streamlit front end in GitHub Codespaces that takes speech, transcribes it, passes text to an LLM to make a song, and fans that output into images, music, audio and short video; it uses Replicate + Groq APIs and models like Llama 3 and musicgen. Requirements show up (GitHub account, Codespaces, API keys). Demo mostly succeeds but some models (Bark) are slow and need retries while F5 runs faster — speech→LLM→audio/image/video via Streamlit in Codespaces.
AI-written summary · about
chapters (10)
- 0:00Workshop introduction
- 1:47Multimodal Gen AI demo overview
- 10:55Why composable AI systems
- 23:13Data science to software shift
- 39:53Generating speech with AI
- 47:04Chaining text to speech and LLMs
- 57:26Text to music and images
- 1:06:07Building frontend
- 1:17:44Deploying and testing
- 1:28:19Q&A
related talks
40mCreating Interactive GenAI applications powered by Streamlit, open source LLMs
34mUsing Appsmith for building GenerativeAI apps
36mGenerative AI + Python: Unlocking Efficiency, Personalization, and Insight
42mEffortlessly Build, Run, Secure, and Manage Traffic for a Generative AI Application From...
37mBuilding Multimodal AI Agents From Scratch
01h 34mHands-on AI workshop: Multimodal agents
40mThe Marvelous Magic of Multimodal AI
41mFrom RAGs to riches: Build an AI document interrogation app in 30 mins