Kestrel
Low, weary swordswoman who has already lost this war once
Multi-voice audio drama workbench
Turn any script into a full-cast audio drama
Paste a scene. CastDub works out who is speaking, gives every character their own voice, lets you direct the emotion line by line, and exports a finished mix. Not a single narrator reading everyone.
40voices in the library
20voice categories
8ready-made workflows
4steps from script to mix
The difference, immediately
One scene, four voices, no assembly
This is a scene as CastDub receives it and as it comes back. Two characters and a narrator, each with their own performer — the thing a single-voice tool cannot give you no matter how good the voice is.
MARIA: You said you'd be here at six.
DANIEL: I said I'd try. There's a difference, and you know there is.
MARIA: She waited by the window the whole time.
DANIEL: …Don't.
NARRATOR: Outside, the last bus pulled away from the stop without stopping.
Ashford
Velvet-voiced villain who apologises before he ruins you
Hale
Documentary narrator with a hand on the listener's shoulder
Cass
Seventeen and furious about it, in a good way
Samples on this page are placeholders while the public demo audio is produced. Every voice in the library has its own card on the voices page.
Four steps
Import → auto-cast → adjust emotion → export
- Import your script
Paste it or upload a text file. Screenplay format, Name: lines and ordinary prose with dialogue tags all work — there is no template to fill in first.
- Auto-cast the scene
CastDub finds every speaker and proposes a voice for each, based on how you described them. Audition alternatives back to back on the same line and keep what fits.
- Direct the emotion
Set a resting register per character, then mark the specific lines where it breaks. Restraint is what makes the turn land — direct five lines, not fifty.
- Export the mix
A mixed MP3 or WAV, per-character stems for scoring, and an SRT that is already in sync because it came from your script rather than from a transcript.
What people make
Pick the shape of your project
Fanfic audio drama
Dialogue-heavy, character-driven, and the worst thing to hand to a single narrator.
Children's stories
A warm adult narrator, child voices for dialogue, and no child in the recording pipeline.
Short drama dubbing
Eighty episodes, a consistent lead voice, and captions per episode from the same script.
OC, VTuber & visual novel
Hundreds of branching lines that must still sound the same on a second playthrough.
Audiobooks & audio drama
One narrator for the prose, a separate voice for every character in it.
Voice assets
A character card is the unit you reuse
Consistency in a series never fails because of the model. It fails because a person re-picks a similar-but-different voice six weeks later, and listeners notice immediately — human hearing is tuned for exactly that.
A character card holds the name, the persona, the voice, the resting emotion and any pronunciation fixes for that character's vocabulary. Open a new project, drop the card in, and episode twelve sounds like episode one without anyone remembering anything.
Cards travel across projects, which matters most for the things people actually produce: a serialised drama, a campaign that runs for two years, a visual novel where players compare routes.
Aria
Bright shonen-anime lead who narrates her own fight scenes
Kestrel
Low, weary swordswoman who has already lost this war once
Milo
Nine-year-old with a loose tooth and a very large opinion
Wren
Bedtime-story narrator, slow and warm, never wakes anyone up
Reviews
We have not shipped long enough to have testimonials
Every launch page you have ever read has five glowing quotes on it. We do not have real ones yet, and inventing them is the single most common thing that makes a new product untrustworthy. When there are quotes here they will be from named people who actually used it, with a link to what they made.
Pricing
Free to start, and the free quota renews
The paywall is emotion control, cast size and clone slots — not the voice library. Everyone gets every voice.
Free
$0 / month
Enough to finish a short scene and hear what your script sounds like cast.
- 10,000 characters every month — the quota resets, it is not a one-time trial
- 2 cast members per project
- 5 minutes of finished audio
- Exports carry a short CastDub watermark
Creator
$12 / month
For one person turning their own stories, fanfic or scripts into finished audio.
- 5 cast members per project
- 60 minutes of finished audio a month
- No watermark, commercial license included
- 1 voice clone slot
- SRT subtitle export
Studio
$29 / month
For serialised work: long-running dramas, audiobooks, dubbing pipelines.
- Unlimited cast members per project
- 200 minutes of finished audio a month
- Line-by-line emotion control
- Multitrack and stem export (one WAV per character)
- 3 voice clone slots
Team
$69 / month
For a studio where a writer, a director and an editor touch the same project.
- 5 seats, shared projects and shared character cards
- 500 minutes of finished audio a month
- 10 voice clone slots
- Priority rendering queue
Frequently asked questions
What is an AI audio drama maker?
It is a tool that takes a script with several speakers and produces performed audio with a different voice for each of them. That is a different job from text to speech, which converts a block of text with one voice. The work here is casting, directing each line, keeping characters consistent across scenes, and exporting something you can mix.
Can I use different voices for each character?
That is the entire point. CastDub detects who is speaking, assigns a distinct voice per character, and keeps that assignment stable across the project so the same character sounds the same in scene one and scene thirty.
Can I use the audio commercially?
Yes on any paid plan — Creator, Studio and Team all include a commercial licence and export without a watermark. The free plan is for personal and non-commercial use and adds a short audio watermark. Your script remains yours on every plan.
How many characters can one project have?
Two on Free, five on Creator, unlimited on Studio and Team. Most single-creator scenes land between three and six speakers including a narrator, so Creator covers a lot of work and Studio covers ensemble pieces.
Do you support voice cloning, and whose voice can I clone?
Cloning is available on paid plans, and only for a voice you are entitled to clone: your own, or someone who has given recorded consent. We do not clone public figures, performers, or anyone who has not consented, and we never clone the voice of a minor under any circumstances. The full rules are on our voice cloning policy page.
Which languages are supported?
English today, across American, British and Australian accents, with the full library in every age band. Japanese, Spanish, Brazilian Portuguese and Korean are next — each language page says plainly what exists and what does not, rather than listing voices that are not there.
Can I export subtitles?
Yes, on paid plans. Because the subtitle file is generated from your script rather than transcribed from the audio, names and invented words are spelled correctly and the timings are already in sync.
Is there a free plan?
Yes, and its quota resets every month rather than being a one-time trial. Free gives you two cast members, five minutes of finished audio and 10,000 characters a month, which is enough to produce a complete short scene and decide whether the tool fits how you work.
Consent and ethics
Whose voice you are allowed to use
A cloned voice needs the person's consent, recorded
Before we build a voice model we require a spoken consent statement from the person whose voice it is. No consent statement, no model. This applies to your own voice too — it is the same recording step, and it means every model in the system has a provenance record attached to it.
We do not clone public figures or performers
Not as parody, not for research, not with a disclaimer. A synthetic recording of an identifiable person is indistinguishable from a real one once it leaves the project, and there is no framing that makes that safe. Accounts that attempt it are closed.
We never clone a minor's voice
There is no plan, agreement or parental permission that unlocks this. The child character voices in the library are synthetic models built for the purpose, not recordings of children, which is precisely why several children's publishers prefer them.
Say when audio is generated
We are not in a position to enforce this, and we will say it anyway: audiences respond well to disclosed synthetic audio and very badly to discovering it afterwards. It costs one line in a description.
Cast your first scene tonight
Paste a script, let CastDub assign a voice to every character, and export a finished drama.
Create for freeNo credit card. Free plan renews every month.