How Much Does It Cost to Make an Audio Drama?

Most cost questions about audio have one narrator in them. An audiobook is one voice for ten hours. A podcast is one or two hosts and an edit. An audio drama is neither, and the reason it costs more is not that the hourly rates are higher -- they are roughly the same people. It is that an audio drama buys the same hours several times over, and then buys jobs that a solo recording never needed at all.

This is a breakdown of where that money actually goes, which parts scale with cast size, which parts scale with runtime, and which parts do not go away when you switch to AI voices. If what you are costing is a single-narrator project, the per-finished-hour explainer covers the unit itself, and the podcast cost breakdown covers the two-host case. The full article index has the rest.

Why a drama costs more than an audiobook of the same length

A finished hour of audio drama and a finished hour of audiobook are the same amount of listening. They are not the same amount of work, and the gap is structural rather than a matter of ambition.

An audiobook is one performer reading continuously. The recording time is roughly the reading time plus retakes, and the edit is one track. An audio drama of the same length has six performers who each record only their own lines, which means six separate sessions, six separate edits, and a mix that has to place all six in the same room. The listener hears one hour. Production touched it many more times than that.

The cast is the line item that multiplies

Voice work is usually quoted one of two ways: per finished hour of delivered audio, or per session. Published rates for professional narration commonly land somewhere in the low hundreds of dollars per finished hour and run several times that for experienced or union talent -- treat those as ranges, because they move with region, experience and exclusivity.

The rate is not where a drama budget gets away from you. Cast size is. A script with six speaking parts does not cost six times a narration rate in a simple way either, because most of those parts are small: a character with forty lines is not a finished hour of anything. What it does cost is six minimum session fees, six scheduling windows, six sets of pickup requests when a scene changes, and six people who have to be available again in three months when episode four needs them.

The expensive thing about a cast is rarely the biggest role. It is the long tail of small ones, each carrying a floor cost unrelated to how much audio it produced.

The jobs that only exist because there is a cast

A solo narration has a narrator and an editor. A drama adds roles that a one-voice project genuinely does not need, and they are not optional extras.

Someone has to direct. When six performers record separately, nobody hears the scene they are in, and without a director holding the shape of it you get six technically clean reads that do not sound like a conversation. Someone has to do sound design -- decide what the room is, where the door is, what is happening outside. And someone has to mix: match levels and tone across performers recorded in different spaces on different gear so they sound co-located.

Those three jobs are commonly quoted per finished minute or per episode rather than per hour, and they are why a drama budget keeps growing after recording ends. On a dense script, post can rival the cast.

Sound design is a per-scene cost, not a one-time one

The thing that surprises people costing their first drama is that sound design does not amortize. A music bed can be reused. A room tone can be reused. But a storm that has to build behind a specific argument, or footsteps that have to arrive on a specific line, is built for that scene and only that scene.

So the honest way to budget sound design is by scene count, not by runtime. Two episodes of the same length can differ by a factor of several if one is two people talking in a kitchen and the other is a chase through a market. Count your scene changes and your effect cues before you price this, not your minutes.

The licensing bill people forget

Music and effects are the line item that shows up after the budget is approved. Library music and sound effects are licensed, and the license terms matter more than the sticker price: whether it covers commercial use, whether it covers distribution on the platform you are actually shipping to, whether it is per-track or a subscription, and whether it survives you monetizing the show later.

An original score is a separate commission. A commercial library subscription is a recurring cost that has to stay current for as long as the show is distributed. Neither number is universal enough to quote, but both are easy to leave out of a spreadsheet and hard to remove once a season is scored.

What the AI-assisted path actually changes

Generated voices change the shape of the budget rather than shaving a percentage off it. The cast stops being a multiplier. Every speaking character gets its own voice from the library, or a cloned one, and adding a seventh character costs the words that character speaks -- not a seventh session fee, a seventh schedule, and a seventh availability risk three months from now.

Scheduling disappears entirely, and with it the cost that never appears as a line item: the weeks a project sits still because two performers cannot be in the same week. Pickups stop being a session. Changing a line in episode two after episode five is written is an edit, not a recall.

Sound design moves too. Auto-Assign Sounds reads the scene and places music, soundscapes and one-shot effects to match what is happening, which turns the blank-page part of sound design into a review pass. It is a starting point rather than an answer -- you still decide what stays -- but the labor it replaces is the tedious half.

In this model the meter is words per month. The free account is 1,200 words a month with no expiry, which is a scene rather than an episode, and the paid tiers scale the same single number upward. That is the whole pricing surface: how much audio you generate, not how many characters are in it.

What is still not free

Four things survive the switch, and pretending otherwise is how AI-assisted projects go over budget in time instead of money.

Writing the script is untouched. A drama is dialogue carrying everything that prose narration used to carry, and no tool writes that for you. Casting is still a decision -- the library has over a hundred voices to browse and preview, and picking six that sound different from each other in the same scene is craft, not search. Listening is still mandatory: every scene, end to end, at least once, which for a season is real hours on a calendar.

And direction moves rather than vanishes. With human performers you direct in a booth. With generated ones you direct in the text: per-line emotion tags, pauses at four levels -- inline, project-wide paragraph default, per-paragraph override, and intro pauses -- and the pacing decisions that used to be a note to an actor. It is genuinely faster. It is not nothing, and a budget that assumes it is nothing will be wrong about the calendar even when it is right about the money.

Budgeting a scene, an episode, a season

Cost a drama from the bottom up, in that order, because each level exposes a different mistake.

Price one scene first. It tells you your real per-minute post cost, which is the number most budgets guess at. Then price an episode by multiplying scene cost by scene count and adding the fixed per-episode items -- the mix pass, the intro and outro, the final listen. Then price a season by adding what only appears at season length: cast continuity across months, a music license that has to stay live, and the storage and delivery of a growing library of finished chapters.

A season budget built by multiplying one episode by ten is wrong in a predictable direction. It undercounts continuity and it undercounts licensing, which are exactly the two costs that arrive late.

The cheapest way to find out what your show costs is to make one scene of it and listen. Take the most demanding scene you have -- the one with four characters and a storm -- and build it end to end before you commit to a season. The free account covers about a scene, no credit card. Start a project and get a real number instead of an estimate.