What “The Brutalist” actually did with AI (And How You can do Too)

Think about the sound of The Brutalist. How do you create the audio for a world that doesn't exist? For the film’s sound team, part of the answer involved AI - specifically Respeecher, used narrowly to refine the accuracy of Hungarian dialogue after it was recorded, preserving the actors’ performances rather than replacing them.

They did not use Respeecher to turn text descriptions into general sound effects; that kind of text-to-audio generation is what tools like Stable Audio are built for. 

This guide will walk you through building your first complete sound scene using nothing but your browser and a few well-chosen words. It’s not a magic trick. It's a real, professional workflow made accessible. Forget the hype and the confusing jargon. Let’s build something right now.

Step 1: See the Big Picture First

Before you click a single button, here’s the professional process you’re about to complete, translated for the AI era:

Traditional Step	Your AI Equivalent	What You’ll Learn
Record ambience	Describe it in Stable Audio	Shape space through language
Record Foley	Generate specific actions	Add a focal sound for realism
Mix & balance	Blend tracks in Soundtrap	Make your world feel believable
Master	Export directly from Soundtrap	Produce a polished, shareable clip

That’s the entire map. Four stages, two free tools, one clear outcome. By the end, you’ll have your first finished sound scene.

Step 2: Generate the Ambience

Visit StableAudio.com and type your first prompt:

“quiet apartment room, faint fridge hum, soft traffic outside window, 15 seconds.”

Click Generate and listen. You’ve just created a room tone, the invisible audio glue that makes an environment feel real.

🎧 Heads Up: The free version of Stable Audio is for personal, non-commercial use only. It’s perfect for learning, but if you’re using this for YouTube or any monetized project, you’ll need a paid license.

💡 Tip: Run the prompt twice. Each render sounds a little different. Layering these variations later adds a natural, lived-in feel to your scene.

Step 3: Add the Moment

Now give your scene a heartbeat. Prompt again:

“hand placing ceramic mug on wooden table, natural echo, 2 seconds.”

This one sound tells a story. Without it, the ambience is just background. With it, the space feels inhabited. Download both files, your ambience and your mug sound, and move to the next stage.

Step 4: Mix and Balance

Head to Soundtrap.com and create a new project.

Import both of your audio clips.

Align them so the mug sound lands where it feels natural. Real timing isn’t perfect, so give it a small offset from the visual action.

Adjust the volume: pull down the ambience until you feel it more than you hear it.

Add short fades to smooth the start and end of each clip. Hover the region and drag the white fade anchors in the top corners inward.

Play it back. If muting the ambience makes the space suddenly feel empty, you’ve nailed the balance.

Step 5: Export Your Scene

That's it. Hit the Export button in Soundtrap.

If you need WAV/high-quality masters (typical for broadcast/strict delivery specs), note that WAV downloads are only available on specific paid Soundtrap plans. Free tiers usually export lossy formats.

Step 6: Refine with Prompts

AI sound isn’t always perfect, but you can fix almost anything by describing a solution.

Too clean? Go back to Stable Audio and generate a new prompt for “soft white noise hiss, low volume, 10 seconds.” Layer it quietly under your ambience to add air.

Too thin? Duplicate your ambience track and nudge one copy a few milliseconds forward. This tiny echo creates instant stereo depth.

Timing off? Trim or slide clips manually until they breathe with your scene.

Weird glitch? Keep it. Sometimes AI’s mistakes become your most unique textures.

Step 7: Keep Building

Now that you understand the flow, try variations:

“An old elevator door opening with a metallic groan, narrow hallway.”

“In a forest at dawn, birds chirping, light wind through leaves.”

“A distant thunderstorm rolling over an open field.”

Every prompt trains your ear. Every render teaches you how AI hears physics. You’ll start noticing how words like “muffled,” “hollow,” “sharp,” or “narrow” shape sound. That’s how you evolve from using AI to designing with it.

The Takeaway

You no longer need permission, a budget, or a studio to create the sounds in your head.

You just completed a professional workflow using free, browser-based tools. From now on, every sound you imagine is only a sentence away.

Go write it.