What You Actually Need to Start · Volume Two of the Music That Listens series

By Ted Kocher

For anyone who ever wondered what happens after the music is written.

© 2026 Ted Kocher. All rights reserved.

Foreword

You have the music — yours, or a composer's, or a classroom project's. This book is about everything that has to happen next for a game to actually play it, and it is written so that anyone can read it. If you can follow a recipe, you can follow these five steps. No studio required, no jargon unexplained.

The book reads at three depths, and you choose:

STUDENT is the book itself — the five steps, the ideas, and the reasons, in plain language. PRO folds in the working detail: the disciplines, numbers, and habits of people doing this daily. EXPERT is the deep layer: the full manual craft — session builds, comping, editorial reverb, archival ritual — written down deliberately, because music editorial is becoming a lost manual art, and somebody should write it down before it goes.

One idea governs all five steps, and it appears at the very end: paper score, bleeding-edge everything else. The best tool is sometimes four hundred years old and sometimes four months old, and the craft is knowing which is which.

About This Book

Five steps: Assets (turning music into parts a game can use), Systems (the promises that keep a production findable for years), Never Lose an Asset (exactly what it says), Infrastructure (wires, networks, speakers), and Session Preparation (the day with real players in a real room, and everything around it). Each step ends with a short check — answer it right there in the book. Instructors: the checks are the first mile of the certification path.

The Music That Listens Series

Volume One — The Student Handbook. The Art and Systems of Game Music Design. The ideas: why game music listens, how adaptive systems are designed, and how a design becomes a shipped behavior. Everything in it can be done in a browser.

Volume Two — The Technical Manual. What You Actually Need to Start. Five real steps — assets, systems, never losing anything, infrastructure, session preparation — written so anyone can read them, with PRO and EXPERT layers that go exactly as deep as the reader does. The deep layer preserves the manual art of music editorial, on purpose, before it's gone.

Volume Three — The Teacher's Manual. Courses, Rubrics, and the Instructor's Craft. Three complete course formats — a one-day workshop, a seven-week intensive, and a fifteen-week semester — with the workbook, rubrics, and everything a non-author instructor needs to teach them.

Table of Contents

• Step 1: Assets
• Step 2: Systems
• Step 3: Never Lose an Asset
• Step 4: Infrastructure
• Step 5: Session Preparation

Step 1Assets

You have the music. It is sitting in your project as one beautiful finished mix, and you are proud of it, and it is almost useless.

Here is why. A game cannot play your mix. A game plays parts. It raises the strings when the player is losing. It drops the drums when a character speaks. It takes your piece apart and reassembles it, differently, for every single player. So the first real step is turning the music you have into parts a game can use. Those parts are called assets, and the most important assets are stems — your music, delivered as separated layers instead of one finished whole.

Wide stems

Deliver stems, and deliver them wide — more layers, not fewer. The drums on their own. The melody on its own. The pads, the bass, the weird little details: separate. When in doubt, split. Nobody in the history of interactive music has ever complained that the stems were too flexible. Every separate layer is another handle the game can grab — one more thing it can bring in, drop out, or turn down at exactly the right moment. Your job at this step is to hand the machine as many honest handles as the music contains.

Early and often

Deliver assets the moment they are useful, which is long before they feel finished. The build needs music the way a rehearsal needs a pianist: now, imperfect, replaceable. A rough loop in the game this week teaches you more than a polished loop next month, because the game is where you find out what the music actually does when a player is standing inside it.

Provisional mixes are not just acceptable — they are the bloodstream of a production. Just tell the truth about them: name them _temp, and everyone who touches the file knows exactly what they are holding. The suffix is a promise that a better version is coming. Keep the promise.

The file tells you what it is

Every file you make should be self-identifying: anyone can pick it up cold and know who made it, what it is, and where it belongs — from the name alone, without opening it, without asking you, ideally without ever having met you.

Here is a real file name, lightly disguised, from a real production:

final_v6_unmixed_mastered_DO_NOT_USE_use_this_one_v2_WIP.wav

Study it. Every word was added by a reasonable person having a reasonable day. It is final, and it is version six. It is unmixed, and also mastered. It must not be used, and it is the one to use. It is, in closing, a work in progress. This file shipped nothing and destroyed several afternoons, and somewhere on a server it still exists, radiating uncertainty.

The cure is a name that states what the thing is, never how it came to be. _final_FINAL_fix2 is a diary entry. Diaries are for feelings.

Fail small so it lands

You will get bad deliveries — from collaborators, from young composers, from yourself in March. That's part of it. Don't write a delivery spec nobody will read. Let the first mistake happen while it is cheap, then teach from the actual mess. Correcting a bad delivery once costs an hour. Correcting it every time costs the project. A lesson learned from your own wreckage sticks; a lesson learned from a PDF does not, because nobody reads the PDF. Fail small so it lands.

The test of a good delivery is silence: if nobody ever has to ask you a question about it, you delivered.

The step: turn the music into wide, honest, self-identifying stems, and start delivering before you feel ready.

Step 2Systems

There is no universal system for organizing game music. No committee is coming. The naming scheme used on one production would confuse another, and both would horrify a third, and all three ship great games. This bothers people who want the industry to hand them a standard. It should liberate them instead, because the real rule is much simpler:

Build your own system. The point is that yours exists, and holds.

Yours, and it holds

A system is a set of promises you make to your future self: where things live, what they're called, which copy is the truth. The content of the promises matters less than the keeping of them. A mediocre convention followed for five years beats a brilliant one abandoned in October. When a team already has a convention, adopt it completely — the value is the consistency; the syntax is just the costume. When there is no convention, write yours down on day one, and then — this is the entire trick — obey it on the days it is inconvenient, because those are the only days it was ever for.

Two requirements are not negotiable, and both are about time:

It survives years. The calls come later than you think — the remaster, the sequel, the "can we get the stems from the temple level" email, sent four years after you archived the drive. Your system is not for this week. It is a letter to a stranger, and the stranger is you, older, with no memory of any of this.

It is searchable. If finding a thing requires remembering where you put it, you don't have a system — you have a memory palace, and memory palaces have a single point of failure who sometimes goes on vacation. Names, tags, a spreadsheet, a database: anything, as long as the answer to "where is it" is a search and not an archaeology season.

The honest trade

Building the system while you work is ten times more painful right now, and a thousand times easier later. That is the entire economics of this chapter, stated without decoration. Ten times more painful now. A thousand times easier later. And a thousand times easier is a thousand times easier.

Plenty of tools exist to help — databases, batch renamers, templates, scripts. Use them. Build your own when the shelf doesn't have it. Which tool is a personal decision, like a chair; what matters is that you sit in it every day.

The step: write down your promises, make them searchable, keep them on the inconvenient days.

Step 3Never Lose an Asset

Everything in the first two steps assumed your files continue to exist. This is not guaranteed. Hard drives are consumables — all of them are dying, at different speeds, and they do not announce their plans. This step is titled exactly what it means, and it means never. A production can survive bad music. It cannot survive the good take being gone.

Three copies

Triple backups on everything. The copy you work from, a second copy on separate hardware, and a third copy somewhere else entirely — cloud, offsite, a drive at a relative's house, anywhere the same disaster can't reach all three. Two copies feels responsible and is one failure away from being one copy, which is one failure away from being a story you tell at conferences, quietly.

Correct sample rates

Pick the project's sample rate once, on purpose, and keep everything at it. A session that quietly mixes 44.1 and 48 doesn't fail loudly — it fails later, as drift, as clicks, as a cue that is mysteriously, expensively wrong in someone else's session, which is the worst kind of wrong because nobody suspects the file. Sample-rate hygiene is free at the start and unpayable at the end.

If you record a lot

If recording is a real part of your life: excellent drives — the boring reliable ones, not the fast exciting ones — backups actually in order, and a second system standing by. A session that can be killed by one interface, one laptop, or one cable is a séance, not a session. The players cost more per hour than the spare rig costs, once.

The step: three copies, one sample rate, and a spare for anything whose death would end the day.

Step 4Infrastructure

This is the shortest step, because the advice is short. It is also the step people skip, because cables are not glamorous — and then a forty-gigabyte session transfer dies at 91% for the third time, and glamour is reevaluated.

Wires

Wi-Fi sucks. It sucks in a specific, treacherous way: it works well enough, most of the time, for small things, which convinces you it will work for the large thing, which it will not, at the worst moment, wirelessly. Ethernet is great. It is a cable that simply does the thing, at the fastest speed you paid for, every time, for years. Run the cable. Run the fastest you can get. Nobody has ever sat in a studio thinking "I regret this reliable, fast connection."

Audio over the network

If there is any recording in your future — any at all — get Dante now. Audio over the network is one of the few technologies in this field that is simply better than what it replaced: any signal, from any room, to any machine, patched in software instead of crawled-for behind a rack. Learn it before the session that requires it, because the session that requires it will not include time to learn it. This is the rare purchase that buys back actual hours of your actual life.

Monitor for the room you answer to

Your speakers are infrastructure too, and the spec that matters is not flatness — it is translation. The mix that counts is the one your approver hears, on their system, in their room, next to their game, where your perfect cue gets stomped flat by a gunshot. Know what your work will be judged on, and audition it there before they do.

The step: run the cable, get Dante before you need it, and monitor on the ears that will judge you.

Step 5Session Preparation

Sooner or later you will put real players in a real room, and here is what nobody tells you about that day: the session is the smallest part of it. The hours on the stage sit on top of an insane amount of preparation, and the ratio is not close — days of prep on either side of every hour of downbeat. This is not a flaw in your process. It is the shape of the work, and the people who look calm in the booth look calm because the calm was manufactured in advance.

Why the prep is the session

Everything on the stage burns money by the minute: the players, the room, the engineer, the clock. Every question answered during the session was answered at the most expensive possible price. So preparation has exactly one job — move every answerable question backward in time, to where answering it was free:

• Parts that are unambiguous. Proofed, numbered, readable at sight. A player squinting is a session paused, times sixty.
• Tempo maps that are final. The click is the skeleton of everything downstream; a tempo map that changes after recording is a machine for turning money into noise.
• Reference mockups the players can hear. The fastest way to explain the music is the music.
• The interactive anatomy, captured on purpose. You are not recording a performance; you are recording parts — intros, loop bodies, endings, stingers, captured as such, seams planned in advance, not carved out of one long heroic take afterward. Walk in knowing exactly which bars serve which purpose.

Rules of the room

Three rules, learned at other people's expense:

• The talkback is a broadcast. Assume every word said in the booth reaches the entire orchestra. No joking, no ambiguity, no thinking out loud. Sixty professionals are forming an opinion of this session through that mic.
• Pushback goes through the conductor. Only the conductor discusses with the players — and a composer should never conduct their own material. The podium requires a neutrality the author of the notes cannot have.
• Calm is a deliverable. Whatever is on fire in the booth stays in the booth. The room plays the way the room feels.

After the downbeat

The session ends and the work continues: takes get chosen, performances get assembled, the best reading of bar nine gets married to the best reading of bar eleven. This is editorial work, and here is the honest truth about it: editorial is becoming a lost manual art. Fewer people learn it every year, the tools automate its easy half, and the hard half — taste, executed by hand — is quietly leaving the building. The expert layer of this book writes it down anyway. Somebody should.

Paper score, bleeding-edge everything else

One last picture, because it is the governing philosophy of this whole book.

On the stage, the score on the stand is paper, and the markings are pencil. Not from nostalgia — from benchmarking. Paper page-turns are faster. Pencil marks are faster. Four centuries of session musicians have optimized that workflow, and it won.

Meanwhile, the laptop next to the paper is doing absurd, futuristic things: editing the previous take, routing audio over the network, running tools that did not exist eighteen months ago.

Hold both mentalities at once. Use the oldest tool wherever the oldest tool is genuinely faster, and the bleeding edge everywhere it actually cuts — and never confuse being modern with being better, or being traditional with being right. The craft is knowing which is which, per task, forever, while the tools keep changing and the paper keeps not needing a firmware update.

The step: move every question backward to where answering it was free, obey the room — and hold both mentalities at once.

About the Author

Ted Kocher is an AAA Lead Music Designer with eighteen years and fifty-plus shipped titles, three Grammy-nominated scores, and five BAFTA wins. He teaches interactive audio at the graduate level and is the creator of ScoreCanvas (scorecanvas.io), the browser-based design surface for adaptive game music. This book was built independently under Back Pocket Music — personal time, personal equipment, no employer resources.

© 2026 Ted Kocher. All rights reserved.