🧠 Speech / Text → ARASAAC Pictograms (HDC)

Convert spoken or written English into ARASAAC pictograms using Hyperdimensional Computing for offline, semantic word-to-pictogram retrieval.

Uses Whisper tiny for speech recognition. Pictogram lookup uses HDC prototype memory built from ~855 core vocabulary pictograms and WordNet synonym injection — no API call per word at inference time.

The similarity score badge on each card shows retrieval confidence (🟢 ≥ 0.15 · 🟠 < 0.15 · ? below threshold).