# Generating a Spatial Scene: How AI Inference Works in Immersive Audio

> See how AI inference turns a prompt and audio analysis into a spatial scene with azimuth, elevation, distance and movement, ready to render.

Web page: https://spatial9.ai/blog/academy-spatial-inference
Author: Spatial9 Team
Published: 2026-10-04. Updated: 2026-10-16.

## Key takeaways
- During training a model adjusts its weights, while during inference the weights stay fixed and GravityLLM applies what it learned to new music.
- Inference combines a written prompt and an audio analysis, outputs a scene description, and a renderer turns it into ADM, IAMF, binaural or Ambisonics.
- A scene description gives each source a role, azimuth, elevation, distance, movement pattern, speed and range, plus a scene-level stage width.
- Run a brief several times when exploring variations, but reuse the same brief language and A/B tracks when consistency across an album matters.

In this lesson: You will follow a single request through AI inference, from a prompt and an audio analysis to a full scene description, then through the renderer to finished formats. You will learn to read every field of a spatial scene, think about variation and consistency, and understand why a human always reviews the result.

In the last lesson, you watched a model learn. Thousands of small corrections slowly turned examples from real sessions into something like taste. Now comes the moment that all of that training was for.

You hand the model a song and a few words about what you want. Seconds later, it hands you back a world: a vocal anchored in front of you, guitars on either side, a pad drifting overhead, and a stage of a certain width. That moment is called inference.

Inference is where AI stops being theory and starts being a creative partner. Understanding it will make you better at briefing the model, faster at spotting problems, and more confident when you decide what to keep.


## What you will learn

- What inference is and how it differs from training.
- The path from prompt and audio analysis to model, scene description, renderer and formats.
- How to read every field in a scene description, from azimuth to movement range.
- Why the same brief can produce different scenes, and when that is a feature.
- Why human review is a required step, not an optional one.

The rest of this lesson is free with a Spatial9 account: https://spatial9.ai/blog/academy-spatial-inference