Work / Creative software / The Theatre of Stories

BABEL

From written narrative to dramatic structure. An original story, seen another way.

Source auditedPublic repository ↗

Contents

What’s in this book?

Three passes: a minute, how it works, then the deep dive. Stop after any one.

  1. Pass 1 Overview 1 minute
    1. What problem does BABEL solve?
  2. Pass 2 How it works 3 minutes
    1. How was it built, and why that way?
    2. What does a story look like after it?
  3. Pass 3 Deep dive 2 minutes
    1. What does the source show?
    2. What are its limits?
    3. What would Ram test next?

Pass 1 of 3

Overview

The problem BABEL sets out to solve. 1 minute

Pass 1 · Overview

What problem does BABEL solve?

A novel and a visual scenario carry the same story with very different structure.

BABEL is a Python/React prototype that transforms prose into structured visual-scenario blocks, with character customization, chapter navigation, glossary tooling and a co-appearance graph.

My contribution

What I built
A Python pipeline that turns prose chapters into typed scene blocks (narration, dialogue, thought and action, each with its speaker), and a React reader that renders those blocks, keeps characters consistent through a glossary, and draws which characters share chapters.
The tradeoff I chose
The story is turned into typed blocks rather than into finished pictures or video. Typed blocks can be checked against the sentence they came from, and the character graph only counts shared chapters instead of guessing at relationships.
Where it stands
The source was reviewed at commit 3970010, but the full application was not run. The review found gaps in how the stages are wired together, so the example on this page is an original story written by hand in BABEL’s format, not a live result.

Pass 2 of 3

How it works

The decisions behind BABEL, and then a short story set out in its intended structure. 3 minutes

Pass 2 · How it works

How was it built, and why that way?

  1. Type the story. Typed blocks give the reader something more precise to render than a wall of text.
  2. Keep characters consistent. Glossary extraction, YAML storage and prompt context carry names and details between chapters.
  3. Count, do not infer. The character graph counts shared chapters. It makes no claim about relationships.
BABEL architectureOpen diagram

The intended path through the prototype, drawn from source review. It is not a claim that every stage is connected end to end.

  1. 01Source

    Prose chapters enter with a source hash, so an output can be tied to its input.

  2. 02Blocks

    Text becomes typed story blocks: narration, dialogue, thought and action, with speaker and tone.

  3. 03Context

    A glossary provides character context. It is glossary context, not verified vector retrieval.

  4. 04Reader

    A React reader renders typed blocks, with character customization and chapter navigation.

A source audit identified orchestration and reliability gaps. The portfolio demo uses original text and a bounded fixture while those integrations are hardened.

Pass 2 · How it works

What does a story look like after it?

An original passage, set out as scene blocks by hand to show the intended structure.

Authored fixture, not a live BABEL result

Source passage

s1 At midnight, Isha returns a borrowed telescope.

s2 The observatory is empty except for Arun, still trying to map a star that vanished before either of them was born.

s3 “You kept the light on,” she says.

s4 “I wanted the sky to know where to find us,” he answers.

  1. scene

    INT. OBSERVATORY, MIDNIGHT

    Source sentence s1
  2. action

    Isha enters, carrying a borrowed telescope.

    Source sentence s1
  3. narration

    Arun works alone, mapping a star that vanished before either of them was born.

    Source sentence s2
  4. dialogueISHA

    You kept the light on.

    Source sentence s3
  5. dialogueARUN

    I wanted the sky to know where to find us.

    Source sentence s4

Characters and text were created for this site. Each scene block links to the source sentence it came from. This shows the intended structure; it was written by hand, not produced by the pipeline.

Pass 3 of 3

Deep dive

What the source code shows, what BABEL cannot do yet, and what comes next. 2 minutes

Pass 3 · Deep dive

What does the source show?

  • BABEL is a Python/React visual-scenario reading and prose-transformation prototype with typed blocks, character customization and chapter navigation.

    Source verifiedLedger

    Source reviewed at commit 3970010; the full app and live transformation were not executed.

  • BABEL includes glossary extraction, YAML storage and prompt context injection.

    Source verifiedLedger

    No verified embedding-based RAG. Default batch glossary wiring and graph schema interoperability have gaps.

  • BABEL implements a character co-appearance graph from chapter speakers and a circular SVG viewer.

    Source verifiedLedger

    Edges count shared chapters, not inferred social relationships. A glossary schema mismatch was reproduced.

Pass 3 · Deep dive

What are its limits?

  • A source audit identified orchestration and reliability gaps. Those integrations need hardening before any live service.
  • The full application and live transformation were not executed during review.
  • This is structured text. BABEL does not generate video.
  • Only original text appears on this site.

Pass 3 · Deep dive

What would Ram test next?

Harden the integration boundaries, then test how faithfully every beat and speaker survives the transformation. A source hash proves which input was used; it does not prove the adaptation is faithful.