True Voice

A calm, child-first visual communication system for recording and understanding autistic children’s emotions.

Comment

Inspiration

What it does — True Voice

True Voice is a child-first visual communication system.

It helps autistic children:

  • express emotions without words
  • record emotional episodes safely
  • share clear signals with caregivers and Doctors

The Product Gap

  • There is no widely adopted platform that enables autistic children to record and track emotions over time.
  • Existing solutions are largely limited to diagnosis or short clinical therapy sessions.
  • Daily, real-world emotional communication remains unsupported.
  • No application currently enables emotion recording through structured, AI-powered visual imagery.
  • There is a lack of tools that provide historical emotional records, meaningful analytics, and visibility into caregiver involvement and guidance.

Why Visual Expression Matters

  • Visual communication reduces reliance on speech during emotionally difficult moments.
  • Consistent imagery creates a safe and predictable interaction environment.
  • Images act as a central communication mechanism, not visual decoration.

True Voice combines:

  • structured visual choices
  • episode-based emotion recording
  • caregiver-aware interpretation

It is supported by AI-powered image generation and structured analytics.

AI-Powered Image Generation

  • AI is used to generate calm, structured support images based on predefined scenarios and reasoning constraints.
  • Image generation follows registered and controlled prompt logic to ensure safety, consistency, and predictability.
  • OpenAI is used at the reasoning and prompt-composition layer, enabling scenario-aware image generation without free-form or unsafe outputs.
  • Generated visuals are intentionally constrained to support emotional clarity rather than novelty.

Episode Analytics and Parent Involvement

  • Each emotional episode is recorded as structured data rather than unstructured text.
  • Past episodes can be reviewed to understand:
    • recurring emotional patterns
    • severity trends
    • context-specific discomfort
  • The system tracks caregiver attendance and guidance, providing visibility into how parents or guardians respond over time.

Supporting Emotional Well-Being and Growth

  • True Voice supports children’s emotional well-being by reducing pressure on verbal expression.
  • Repeated, predictable visual interactions help children feel safer and more confident when expressing discomfort.
  • Over time, recorded episodes and caregiver actions create a clearer picture of:
    • emotional development
    • behavioral patterns
    • caregiver-child interaction quality

Clean and Interpretable Insights

  • Analytics are designed to be simple, interpretable, and caregiver-friendly.
  • Historical data highlights both the child’s emotional behavior and the level of parental involvement.
  • These insights help caregivers respond more effectively during future situations without replacing human judgment.

##What It Does True Voice enables non-verbal emotional check-ins through a calm, step-based visual flow.

The child:

🦸 Selects a familiar hero for comfort and safety 🏠 Chooses context (home, school, outside) 😊 Selects feelings using visual cues 🚦 Indicates severity (mild / severe) without speaking

The system:

🧾 Stores each interaction as a structured episode 🖼️ Displays a calming supportive image 👨‍👩‍👧 Shares context with caregivers (not alerts) 🔍 Beneath the Surface: Pattern Discovery True Voice quietly builds longitudinal emotional context without collecting sensitive medical data.

It can:

📈 Track recurring symptoms + severity trends 📍 Identify contextual patterns (where distress occurs most) 🚨 Differentiate emergency vs non-emergency episodes 🤝 Record caregiver response + attendance

🛠️ How We Built It 🧩 Technical Architecture

True Voice is designed with demo stability, emotional safety, and future extensibility as first-class constraints.

The deployed demo intentionally avoids runtime dependencies on external services, while the architecture remains ready for live generation.

Frontend Design

The frontend is built as a child-first, accessibility-focused system that prioritizes emotional safety, predictability, and low cognitive load.

Interaction Model

The UI follows a strict, step-based flow:

hero → place → symptom → severity

Each step:

  • presents one clear decision at a time
  • avoids branching complexity
  • prevents overwhelming choices

This sequential structure aligns with how many autistic children process information visually and step-by-step.

Accessibility-First Principles

The interface is intentionally designed to reduce stress:

  • predictable transitions
  • limited visual variance
  • consistent component placement
  • no forced verbal input

Animations and colors are kept calm and minimal.

State Management

Structured Emotional Episodes

Episodes are stored as structured objects (not free text), ensuring:

  • clear interpretation
  • deterministic replay
  • unambiguous emotional context

Local Storage (Demo Mode)

For demo and judging:

  • episodes are stored locally in the frontend
  • no backend dependency is required

This enables:

  • reliable demos
  • instant replay of past episodes
  • parent review without network risk
  • analytics without latency or API failure

Image Handling & Reliability

Deterministic Image Mapping

Images are mapped using scenario keys:

hero + place + symptom + severity

This guarantees:

  • familiar visuals across sessions
  • emotional continuity
  • no random or unexpected imagery

Self-Hosted & CDN-Cached Assets

All support images are:

  • self-hosted
  • CDN-cached
  • served reliably

This eliminates throttling, prevents expiration issues, and ensures stable behavior during judging.

Backend (Node + Express)

  • Token-protected APIs
  • Scenario-keyed image service
  • Returns deterministic outputs based on structured input

Core Backend Routes

  • POST /api/auth/login
    Demo authentication (JWT-based)

  • POST /api/support/generate
    Builds a structured prompt and returns a cached or generated support image

Structured Prompt Composition

Frontend selections are converted into a deterministic scenario key:

hero_scenariosymptom_severity

This key is used for:

  • caching and reuse
  • consistent image generation
  • preventing random variation across runs

Image Generation Modes

Judging Mode (Deterministic)

  • Images are pre-generated using FIBO
  • Backend returns static or cached URLs
  • Ensures speed, reliability, and stable demos

Live Mode (Production-Ready)

  • Same endpoint can generate images dynamically using FIBO
  • No frontend changes required

Episode Handling

Current (Demo)

  • Episodes stored in frontend local storage
  • Enables replay, attendance, and analytics

Future

  • Backend persistence per scenario key
  • Long-term analytics and trend tracking

FIBO Integration Highlights

FIBO is used for controlled, child-safe visual generation:

  • scenario-bound prompts (no free-form input)
  • consistent, calming visual style
  • deterministic reuse of familiar imagery
  • reduced sensory overload through bounded variation

🚀 What’s Next

⏳ Short-Term (0–3 Months) Deploy the application end-to-end with live FIBO image generation enabled post-judging

Introduce scenario-based caregiver message personalization

Expand the symptom and context library to cover a wider range of daily situations

🌱 Mid-Term (3–6 Months) Add a secure emergency chat feature for high-stress situations, supported by calming FIBO images for faster and clearer responses

Launch a child–parent relationship dashboard to track emotional understanding and progress over time

Introduce richer stories and situational contexts, co-created with parents and caregivers

Enable parent-assisted visual input (e.g., photos of home, school, or surroundings) to generate more accurate and personalized visual prompts

Develop more structured and expressive prompt frameworks to help children communicate emotions with greater clarity and confidence

💛 Closing True Voice is not a diagnostic tool. True Voice is designed for autistic children who find verbal expression difficult. It offers a calm, visual way to communicate emotions and discomfort without pressure. By recording moments consistently, it helps parents see patterns instead of isolated incidents. This reduces misunderstanding and emotional guesswork at home. Over time, these records create meaningful context for doctors and therapists. Not just what happened once but what happens repeatedly. ** True Voice bridges the communication gap between child, parent, and clinician with empathy.**

Built With

Share this project:

Updates