Free Voice Acting Practice Scripts
Six original scripts, written for this page, free to use and free to record. They are ours, so there is no copyright question and nobody to credit. But a script on its own teaches you nothing. Read the first section before you record a word of them.
Most people practise a script by reading it out loud. That trains one thing: reading out loud. It is the single most common way to spend two years getting no better.
A script is not the exercise. The script is the raw material for the exercise, and the exercise is making choices and then changing them on demand. That is the actual job. Every casting director in this business is screening for the same thing, and it is not a nice voice — it is whether you can take a note and do something different with it on the next take.
So: a method first, then the copy. Six scripts, across the six categories that actually exist as paid work — commercial at thirty and sixty seconds, corporate and e-learning narration, animation, video game, and literary narration. Each one has a spec line at the top, written the way a real client writes a brief, because learning to read a brief is half of learning to book.
How to actually practise with a script
Do this with every script below. It takes about twenty minutes per piece, and it is worth more than fifty cold read-throughs.
One. Read it once, cold, out loud, and accept that you sounded like a person reading. That is your baseline and it is meant to be unimpressive. Two. Find the single idea. Write it in the margin in your own words, in one sentence. If you cannot, you do not understand the copy yet and no amount of vocal warmth will hide that. Three. Answer two questions in writing: who am I talking to, and what do I want them to do in the next ten seconds? Name a real, specific person. Not an audience. A person.
Four. Mark the turn. Almost every piece of copy pivots once — the sentence where the problem becomes the solution, or the joke lands, or the character decides. Put a line through the page there. Five. Put the script face down and paraphrase the whole thing from memory, in your own words, to that specific person. This is the step everyone skips and it is the step that works. Six. Pick the script back up and read it again. It will sound different. That difference is the entire skill.
Seven. Record two more takes, each with exactly one variable changed — slower, closer, take the smile out, land the last line harder. Not vaguely better. One named change. That is what direction sounds like, and you are practising receiving it before anyone is there to give it.
Commercial
Commercial is where the money is, and it is the most technically unforgiving category on this page. The timing is real: a :30 is thirty seconds, not thirty-four. Time yourself. If you are over, the answer is almost never to speak faster — it is to find the two words you were leaning on for no reason.
SPEC: Regional broadcast and digital, :30. Any gender, sounds 30 to 45. Real person, not an announcer — this should sound like someone talking to a friend across a table, not selling. Smile in the voice on the tag, then drop straight into a dry, fast legal read. No hard sell anywhere.
Nobody grows up dreaming about a savings account.
(beat)
You dream about the workshop out back. The spare room with the good light. The van with the bed in it and no fixed address.
At Harbour Row, we do not pretend to know what yours is. We just help you get there sooner.
No minimum balance. No monthly fee. No small print we would be embarrassed to read out loud.
(warm, settle, this is the tag)
Harbour Row Credit Union. Your money. Your plan.
(dry, fast, no colour — this is legal and it has to fit)
Federally insured by NCUA. Terms and conditions at harbour row dot com.
SPEC: National brand film, :60. Sounds 35 to 55. Low, close, unhurried — you are remembering something, not announcing it. Almost no energy at the top. The lift comes at the date and it is small. Do not sell the last line; let it land and get out.
The first thing you fix is never the thing that is broken.
It is a hinge. A tap that drips. A stair that has been complaining about you for eleven years.
(beat)
You fix it on a Saturday. Badly.
Then you fix it again on the Sunday, properly, because you went and looked it up.
And somewhere in that second attempt, something happens. The house stops being a place you live inside. It becomes a thing you are making.
(small lift — small)
Marchmont Hardware has been open since 1946. Four generations. Same corner.
We will not sell you the biggest one. We will sell you the right one, and then we will show you how to use it.
So bring us a photo. Bring us the broken part in a sandwich bag. Bring us the tap, and the eleven years.
(warm, settled, no push)
Marchmont Hardware. We will show you where it goes.
Corporate and e-learning narration
This is the bread-and-butter category, and it is also the category most exposed to synthetic voice — which means the bar has gone up. A flat, competent, informational read is now the thing a machine does for nothing. What a human still sells here is clarity of thought: knowing which word in the sentence carries the meaning, and leaving room around it.
SPEC: Internal training module, warehouse operations. Neutral and authoritative but not cold — you are the person who has actually done this job, not the compliance department. Even pace, clean consonants, no sell whatsoever. Each marked slide is a separate audio file, so land the final word of each one and leave air. Roughly 140 words a minute.
Module Four. Working at height.
Falls remain one of the most common causes of serious injury in warehouse environments. Most of them do not happen at height. They happen on the second or third rung of a ladder, in the ninety seconds when somebody decided they did not need to go and fetch the right equipment.
(slide)
Before any work at height begins, ask three questions. Can this task be done from the ground? If it cannot, is the access equipment right for the job, and has it been inspected within the last twelve months? And finally — does anybody know that you are up there?
(slide)
Ladders are for short-duration, low-risk tasks only. If the job will take longer than fifteen minutes, or if it needs both of your hands, a ladder is the wrong tool. Use a podium step or a mobile tower.
(slide)
Never stand on the top two rungs. Never lean out beyond the stiles. If you cannot reach it, come down and move the ladder. Moving the ladder takes eleven seconds. A shoulder injury takes eleven weeks.
(slide — button prompt, lighter, invitational)
Select Continue to begin the knowledge check.
Character work
Two pieces here, and one warning across both of them. You are not being asked to do a voice. You are being asked to be a person who happens to be drawn. Play the situation honestly and the character arrives on its own; chase the character first and you will get a cartoon, which is a different and much worse thing.
SPEC: Half-hour animated series, pilot. Character: BARNABY QUILL, a field mouse who has read the manual for everything and trusts none of it. Over-prepared, permanently one bad minute from panic, and — underneath it — genuinely brave. Bright and forward. Use your own voice pushed up, not a squeaky voice. Play the fear absolutely straight; the comedy is in the honesty of it, not in the pitch. The last two lines are the audition.
BARNABY: Right. Right! Everybody stay calm, because I am calm, and I am the benchmark.
(reading from a very small notebook)
Step one of the emergency protocol: identify the emergency. (beat) The emergency is that we are currently inside the emergency. Fine. Good. Excellent. Step one complete, we are ahead of schedule.
(the ceiling groans)
— Step two is optional.
(to himself, faster, unravelling)
I told them. I told them about the load-bearing acorn. I made a diagram. Nobody looks at the diagrams, Barnaby, they said. Oh, they will look at the diagrams now, won't they, when we are all — (catching himself) — calm. Calm.
(one small brave breath. Suddenly quiet and completely steady.)
Okay. Hazel. Take the little ones and go left. I will hold the door.
(beat. Terrified. Entirely committed.)
I have read about doors.
SPEC: AAA action RPG. Character: THE WARDEN. Human, sounds 50 to 65. Never raises his voice, never rushes, entirely certain he is the reasonable one in the room. Do not perform villainy — perform reason. The monologue is a single quiet take. The barks are recorded separately at full effort, one take each; be warned that this is the section that ends voice actors' days early, and that vocal-stress work like this is exactly why game sessions are booked and paid by the hour.
THE WARDEN: You keep using the word free, as though you have checked what it costs.
(a step closer)
I have. Every night for thirty-one years, I have counted it out in bodies. Would you like the figure? No. You would not. You want the feeling.
(quiet, almost kind)
I was you, once. I stood in this exact room and I told a tired old man that he had built a cage and called it a city. He did not argue with me. He simply handed me the keys, and asked me to do better.
(beat)
And I have done better. Nobody starves. Nobody screams in the night. The price is that nobody chooses.
(final. Gentle. Absolutely final.)
Take the keys. Use them, even. But you will never put them down again — and that is the only promise I can make you, and the only true thing anyone in this city has ever said to your face.
— BARKS. Separate takes. Full effort. Do not save your voice.
(alert, sharp) Behind you!
(shouted, across distance) Hold the line — hold it!
(light impact, short) Nnh —
(heavy impact, going down) Agh — no —
(exertion, a swing) Hah!
(death, very small, surprised) ...oh.
Long-form narration
Audiobook is the endurance category. The research on this is unambiguous and worth carrying into every practice session: ACX's own documentation says it takes most producers five to seven hours of work to produce one finished hour of audio. So the skill being built here is not the beautiful paragraph. It is the sustainable one — a read you could still be doing, at the same level, four hours from now.
SPEC: Literary fiction, adult, first person. One listener, close, in a quiet room. Do not act it — tell it. There are two voices in the passage; separate them with pace and placement, not with accents. Working pace is roughly 155 to 165 words a minute. Record the whole thing in one pass and resist the urge to punch in.
My mother kept the good scissors in a drawer that did not open.
Not locked. Warped. You had to lift the handle and pull left and say something to it, and she was the only one who ever knew what to say, and I have thought about that more in the last four years than I have thought about almost anything else.
The scissors were for fabric. Cutting paper with them was, in the private taxonomy of our house, a moral failure — ranked somewhere between lying and leaving the immersion heater on.
On the last afternoon I ever spent in that kitchen, she was standing at the drawer with her back to me, and she said, without turning round:
(hers — flatter, drier, older)
Take them, then. You will only come back for them.
And I said: I do not want them.
And she said: No. But you will.
(beat — then back to the narrator, and let the pace settle)
They are in a tin on the shelf above my sink. They do not cut anything any more. I keep them the way you keep a phone number you cannot call.
How to record it, and how to grade yourself
Record dry. No EQ, no compression, no noise reduction, no gate. When you send real work to a real engineer, that is what they want — raw audio, so they can apply their own chain — and sending a heavily mastered file to a broadcast engineer marks you as an amateur before anyone has heard your read. Practise the way you will work.
The technical targets, from ACX's published submission requirements, are the most concrete numbers in this whole subject: RMS between -23dB and -18dB, peak no higher than -3dB, and a noise floor no higher than -60dB RMS. Record thirty seconds of your empty room in silence and look at that last number honestly. An untreated bedroom typically sits at -40 to -50 dBFS, which is not close.
And do not fix it with software. More people fail quality control for over-processing than for under-processing — the beginner hears hiss, panics, and drags noise reduction until the voice sounds like it is underwater, and the human listener at the other end hears the artefacts instantly. Fix the room, not the file.
Now the honest part, which is the reason this chapter exists at all. You cannot grade your own read. You know what you meant, so you hear what you meant. Listening back to yourself tells you about your microphone and your noise floor and almost nothing about your acting. A script plus a recording device plus no ears on it is a very convincing simulation of practice, and thousands of people are doing it tonight, and most of them will still be doing it in three years.
Take these six scripts to a coach, a class, a scene partner, anyone who will give you one specific note and then make you do it again immediately. The note is the training. Everything above it is just the raw material.
Want Will to Coach You Through It?
Reading is one thing. Working 1-on-1 with a working actor who booked Oppenheimer is another.