Storytime: Sleep Sherpa

AI-crafted sleep stories with conversational voice personas, ASMR narration, and social remix powered by Gemini and ElevenLabs

Comment

Inspiration

70M+ Americans struggle with sleep, yet solutions are either clinical (white noise) or passive (podcasts). We built an experience where you chat with a calming voice persona and receive a story personalized just for you — upload a photo of your pet or a cozy scene, and your AI sherpa weaves it into a soothing narrative.

The goal: make bedtime feel magical again — for adults rediscovering rest and for parents seeking safe, screen-time-appropriate stories for their kids.

What it does

Voice Intake — Select one of 5 ASMR personas and talk to them via ElevenLabs Agents

Image Context — Upload a photo analyzed by Vertex AI Vision to seed story themes

Story Generation — Gemini 3 Pro creates a 5-10 min narrative with grounding exercises

ASMR Narration — ElevenLabs TTS streams whisper-style audio with waveform visualization

Library & Remix — Save favorites, browse community stories, remix with different personas

How we built it

Frontend — Expo SDK 54, React Native, TinyBase (offline-first), Reanimated

AI — ElevenLabs Agents + TTS, Vertex AI (Gemini 3 Pro + Vision)

Backend — Firebase Auth, Firestore, Storage, Cloud Functions v2

UX — Latency masking (pre-gen audio), glassmorphism theme, sleep timer

Challenges we ran into

iOS audio session contention — LiveKit + expo-audio fought over AVAudioSession; built explicit playback/voice handoff

Perceived latency — Pre-generated "mask" clips play instantly while AI works

Remix context preservation — Stories store full provenance for chainable remixes

Accomplishments that we're proud of

5 distinct ASMR personas with unique voice profiles

True voice-driven creation (no typing required)

<5s end-to-end story generation + narration

Social remix mechanic for infinite story variations

What we learned

iOS audio sessions require explicit, deterministic handoffs

Latency is perception — fill the wait with meaningful audio

ASMR is an audio design discipline; one loud tap breaks the spell

TinyBase + Firestore = offline speed + cloud durability

What's next for Storytime: Sleep Sherpa

Custom voice cloning, Android release, premium tier

HealthKit sleep tracking integration

Live collaborative storytelling sessions

30+ language intl expansion

Built With

Share this project:

Updates