HeirloomAI Build for a Friend & Family
Model: Qwen 2.5 (Offline)
Build for a Friend Hackathon

Turn Grandpa's Spoken Stories & Memos into an Heirloom Cookbook

Keep your loved one's voice, memories, and secret cooking techniques safe forever without sending a single byte to a big-tech cloud server.

Grandpa's Vintage Recipe Journal
Zero Data Leak • $0 Cloud Cost

Spoken Audio & Voice Memo Input

Record live or upload Grandpa's voice note file
00:00
OR UPLOAD AUDIO FILE

Drag & drop Grandpa's MP3, WAV, or M4A audio memo here

or click to browse files

Open-Source AI Parser Pipeline

Local Inference

Roommate & Loved One Allergy Profiles

Configure safety rules for your housemates

Open AI Safe Substitution Assistant

Local Privacy Guard

Patient Practice Partner

Helps friends practice cooking terms, stories & languages without cloud logging

Practice Partner (Local AI)
Hello! I am your patient cooking and language practice partner. Tell me what recipe you want to make or ask me any question about Grandpa's recipes! How can I help you today?

Why Open Innovation Matters for HeirloomAI

Built for the "Build for a Friend" Hackathon • Demonstrating why open-weight models and browser local execution outperform closed proprietary APIs.

1. Works 100% Offline

Runs on a laptop with zero internet connection. Grandpa can record family recipes at a remote cabin, on an airplane, or off-grid.

2. Total Data Privacy

Keeps sensitive family stories, voice memos, and personal health/allergy profiles strictly off third-party servers.

3. Model Swappability

Seamlessly hot-swap between Qwen 2.5, Llama 3.2, Mistral, or custom fine-tuned weights depending on memory and language.

4. $0 Operational Cost

Zero per-token cloud API bills. Unlimited transcript processing and recipe generation for life at zero recurring cost.

Architecture & Tech Stack

  • Frontend Architecture: Pure Vanilla HTML5, Modern HSL CSS variables, Async Modular ES6 JavaScript (`index.html`, `style.css`, `app.js`).
  • Audio Processing: HTML5 Web Audio API, MediaRecorder API, Frequency Waveform Canvas Renderer.
  • Open AI Engine Bridge: Dual local pipeline supporting REST API endpoints (`Ollama` & `LM Studio`) alongside Browser Local Rule-based Fallback Parser.
  • Speech Synthesis: Web Speech API (`SpeechSynthesisUtterance`) for hands-free audio narration.