Luby

Voice-first AI creates personalized lullabies. Whisper your baby's name, favorites, and loved ones to get a custom song with vocals and cover art.

Comment

Luby - Hackathon Submission

Elevator Pitch

Voice-first AI creates personalized lullabies. Tell us about your baby's name, favorites, and loved ones—get a custom song with vocals and cover art.

Inspiration

Every parent wants to give their child something uniquely theirs. Lullabies have been passed down for generations, but what if you could create one that's truly personal—featuring your baby's name, their favorite things, and the people who love them most? We wanted to make that magic accessible to every family, using AI to turn a simple conversation into a keepsake song.

What it does

Luby lets parents create personalized lullabies through a natural voice conversation. An AI agent asks about your baby—their name, loved ones, favorite things, and what makes them special. Minutes later, you receive a special lullaby with vocals singing your baby's name, accompanied by AI-generated cover art. Each song is stored in your personal library to play anytime.

How we built it

  • ElevenLabs Voice Agent for natural conversation to gather baby details
  • Gemini 2.5 Flash to write personalized lyrics and generate image prompts
  • ElevenLabs Music API to compose lullabies with vocals
  • Imagen 4 to create unique cover art
  • Next.js 15 + React 19 for the frontend
  • Firebase for auth, database, and media storage
  • Firebase App Hosting for deployment

Lyrics and image prompts generate in parallel; cover art generates asynchronously so parents get their audio faster.

Challenges we ran into

  • Safe image generation: AI image models have strict policies around children. We pivoted to generating dreamy, artistic scenes (starlit meadows, sleeping animals) rather than depicting babies.
  • Voice agent state management: Keeping the conversation flow smooth while collecting structured data required careful coordination between the agent and our backend.
  • Generation timing: Balancing user experience with API latency—we made image generation non-blocking so users get their lullaby audio immediately.

Accomplishments that we're proud of

  • A truly voice-first experience—parents never need to type
  • Parallel AI pipelines that minimize wait time
  • Lyrics that naturally incorporate the baby's name and story
  • A polished, responsive UI that works from 160px to any screen size
  • The emotional reaction when parents hear their baby's name sung back to them

What we learned

  • Voice interfaces require different UX thinking—progress indicators and status feedback become critical
  • AI orchestration is about smart parallelization and graceful degradation
  • Content safety policies shape product design in unexpected ways
  • Firebase App Hosting + Next.js Server Actions make full-stack AI apps surprisingly simple to deploy

What's next for Luby

  • Native mobile app with background playback—keeps playing even when the device is locked
  • Playlists to organize lullabies and set a bedtime rotation
  • Shareable playlists so grandparents and caregivers can listen too
  • Smart volume control that gradually fades the music as baby falls asleep
  • Cry detection that gently raises the volume when baby stirs
  • Longer songs with multiple verses
  • Melody selection letting parents choose different musical styles
  • More languages for lullabies in Spanish, French, and beyond
  • Physical keepsakes like QR-coded cards linking to the lullaby

Built With

Share this project:

Updates