GhostRunner: Turning Every Street into a Story

Description and use

GhostRunner is a local‑first audio drama engine that pre‑compiles a route‑specific story, anchors each line to a precise distance, validates landmarks against OpenStreetMap, and plays the narration in real time. It enables a one‑person side‑hustle selling reusable mission packs.

Problem

Running and walking experiences need real‑time audio narration that is spatially accurate, grounded in real landmarks, and latency‑free.

Outcome

GhostRunner delivers a fully compiled, spatially exact audio drama that runs on a laptop with a 4 GB GPU, enabling a one‑person side‑hustle selling reusable mission packs.

Workflow

• Plan route with Valhalla
• Query Overpass for real landmarks
• Compile route model and chainage
• Director LLM generates story bible and beats
• Beat writer LLM writes 3 variants per beat
• Validator checks word count, landmark allowlist, safety rules
• Pre‑render TTS for a rolling window
• During run: GPS + pace classifier → locate on route → trigger check → select variant → play audio

Evidence

GhostRunner plans a route over real streets, finds the real landmarks along it, and writes an audio drama in which your surroundings are the set

Source

It is spatially exact. Every beat is anchored to a distance along the route, so a line about a turn plays before the turn, not after it.

Source

It is grounded. The narrator can only name landmarks that exist in OpenStreetMap along your route. A validator rejects anything else.

Source

It is fast, because nothing is generated while you run. The story is compiled and voiced ahead of you. Playing a line is a lookup, not an inference.

Source