You have a novel with forty named characters and you are trying to work out how many of them need their own voice. Every casting guide you find answers a different question — how to write an audition script, how to brief a narrator, how to evaluate submissions — and then hands you the same constraint: full-cast production costs two to three times what a single narrator costs, so keep the cast small.
That constraint is real when you are hiring people. But it has quietly become the method. Authors do not cast their books; they ration them, and then work backwards to justify whichever number the budget allowed.
Cast size is not a budget output. It is a property of the text. This guide is the method for reading it off your own manuscript — which characters must have a distinct voice, which benefit from one, and which are better left in the narrator's hands even when a voice is free.
Why the usual answer is a price, not a method
Look at how the question gets answered today and the shape is consistent.
On ACX, Amazon's audiobook marketplace, you post an audition script and narrators submit readings. It is built around casting a narrator — one contract, one royalty split, one person. Voices by INaudio — the platform that operated as Findaway Voices until its founders reacquired it from Spotify in August 2025 — offers a curated roster you can browse and invite. Both are marketplaces for hiring humans, and hiring humans is priced per human.
So the guidance follows the money. ScribeCount's human-narration guide describes full cast as the format that sounds distinctive and markets well but where costs double or triple, and points authors toward duet narration — two narrators alternating point-of-view chapters — as the affordable middle. That is honest advice about a real market. It is also why the median indie answer to "how many voices?" is "one, or two if the book is romance."
The ceiling case shows how far the format can go when money is not the binding constraint: Audible's Harry Potter full-cast productions use more than 200 voice actors across the seven books. Nobody arrived at 200 by asking how few they could get away with.
Between one and two hundred is a decision nobody is teaching you how to make.
Ready to try it on your own book?
Start your first chapter free →Count scenes, not characters
Here is the reframe that does most of the work: a distinct voice exists to tell two speakers apart, so what matters is who shares a scene, not who exists in the book.
Your antagonist in Part One and the harbourmaster in Part Three never speak to each other. A listener has no way to compare them and no reason to. They can share a voice — genuinely share it, not just sound similar — and the production loses nothing. Two characters who trade lines for six pages cannot.
So the real count is not "how many named characters do I have." It is: what is the largest number of speakers who hold a scene together? Walk your manuscript and find the most crowded conversation in the book — the dinner table, the council, the argument in the car. That number, plus the narrator, is the floor of your cast. Everything above it is choice.
For most novels, the crowded-scene number is smaller than authors expect. Fiction is overwhelmingly written in twos and threes, because that is what dialogue can carry. A forty-character epic usually has four or five voices that ever need to be simultaneously distinguishable.
The three tiers of your cast
Once you are counting speakers rather than characters, every character in the book sorts into one of three tiers.
Tier 1 — must be distinct. Characters who hold dialogue-heavy scenes with each other, especially unattributed exchanges. If your text runs long stretches of "..." / "..." without a "she said," the voice is the attribution. Get this wrong and the listener rewinds. This tier is typically three to six people: your POV characters, the antagonist, and whoever the protagonist argues with most.
Tier 2 — benefit from distinct. Recurring characters with real presence but who mostly appear in scenes anchored by a Tier 1 character. A distinct voice here adds texture and makes the world feel populated. It is upside, not necessity.
Tier 3 — leave with the narrator. One-line functional characters: the waiter, the dispatcher, the crowd. A narrator shading a single line is the correct treatment. Handing these a full separate voice is the most common way a production starts to sound busy rather than rich — the audio equivalent of a page where every character speaks in a different font.
The failure mode worth naming: casting every named character is not the maximal version of good casting. It flattens the hierarchy your prose already built. Tier 3 staying in the narrator's hands is what makes Tier 1 land.
What each cast size actually fits
| Approach | Voices | What it fits | Where it strains |
|---|---|---|---|
| Single narrator | 1 | Nonfiction, memoir, first-person literary fiction with sparse dialogue | Crowded scenes; unattributed dialogue; multi-POV structures |
| Duet | 2 | Dual-POV romance, alternating-chapter structures | Ensemble scenes where both POV characters are present with others |
| Small ensemble | 3–6 | Most novels — the Tier 1 count for the majority of fiction | Epics with several simultaneous storylines and large councils |
| Full cast | 6+ | Multi-POV epic fantasy and sci-fi, ensemble thrillers, dramatized productions | Nothing structural — historically only the budget |
Notice that the right-hand column for full cast is empty of craft objections. The constraint on large casts has always been economic, which is exactly why the constraint is worth re-examining when the economics change.
When cast size stops being priced per actor
This is where the method becomes usable rather than theoretical. On a per-actor market, a six-voice production costs meaningfully more than a three-voice one, so the tier sorting above collapses back into a budget exercise.
Midsummerr prices by the word, not by the cast. Full Cast is $3.75 per 1,000 words and Full Production — full cast plus original score and sound design — is $5 per 1,000 words, whether your book needs four voices or forty. A 90,000-word novel produced full cast is about $337; the same novel with a cast twice the size is the same $337.
What that changes is not mainly the price. It is who makes the casting decision and when. Cast size becomes an editorial judgement you can revise after hearing it, rather than a contract you sign before hearing anything. You cast your Tier 1 speakers, listen to the crowded scene that worried you, and promote a Tier 2 character if the scene is still muddy — without renegotiating anything.
You can hear what a sorted cast sounds like in our production of Frankenstein, where the framing narrator, Victor, and the creature each hold their own register through a nested-narration structure that is genuinely hard to follow in single-voice audio. For the format primer, start with our full-cast audiobook guide; for the one-versus-many decision itself, see full cast vs single narrator. If you want help choosing the actual voices once you know how many you need, choosing character voices covers casting to character rather than to demographic.
FAQ
How many voices does a typical novel need?
Three to six voices that must be genuinely distinct, plus the narrator. That is the Tier 1 count for most fiction — POV characters, the antagonist, and the protagonist's main dialogue partner. Books with several simultaneous storylines run higher; first-person literary fiction with sparse dialogue often needs only the narrator.
Can two characters share the same voice?
Yes, if they never share a scene. A listener can only compare voices they hear in proximity, so characters separated by hundreds of pages can share one without the production sounding thin. The rule to protect is that speakers who trade lines must be distinguishable.
Does every named character need their own voice?
No, and casting that way usually hurts. One-line functional characters are best handled by the narrator shading a line. Giving every named character a separate voice flattens the hierarchy your prose already established and makes the audio sound busy rather than populated.
Is a full cast worth it if my book is mostly narration?
Often not. If your book carries long stretches of interior narration with sparse dialogue — much literary fiction, most memoir — the extra voices have little to do. Cast size should follow dialogue density, which is why the crowded-scene count is the better test than the character list.
Can I change the cast after hearing the audiobook?
On a per-actor hire, changing the cast means re-contracting and re-recording. On Midsummerr, casting is an editable production decision — you can hear the book, then promote or merge voices and regenerate the affected chapters. That is what makes the tier method practical rather than something you have to guess right up front.
The short version
Stop asking how many voices you can afford and start asking how many your busiest scene requires. Sort your characters into the three tiers, cast Tier 1 properly, leave Tier 3 with the narrator, and treat Tier 2 as the dial you turn after listening. The number you arrive at is a reading of your own book rather than a reading of a price list.
Produce your first 5,000 words free and hear your crowded scene before you commit to a cast.
