Audio Drama Maker
Turn any chapter into a full-cast audio drama
Paste the text. Characters are detected, voices cast, every line performed with emotion — then adapt it into Spanish, Portuguese, Japanese and more with the same cast. Finding the cast and previewing need credits in your account but use none; renders are billed by length.
Automatic casting
Dialogue, speech verbs and forms of address decide who says each line; the narrator and the characters are performed separately.
Every line directed
Sneers, sobs, whispers and shouts are acted, not read.
Subtitles and labelling
Every line is timed; export SRT for your editor. Files carry an AI-generated label.
- 1Paste a story
- 2Check the cast and script
- 3Preview and render
Finding the cast needs credits in your account but uses none. Rendering starts at $3.90: one-time, no subscription.
Who uses the audio drama maker
Story-time videos
A narrator plus a voice for every character, with the emotion in place. Viewers stay to the end.
Animated comics and shorts
Paste the storyboard lines and get every character at once, with subtitle timing.
Audiobooks and radio plays
Whole chapters cast automatically, one cast through the book, at a fraction of studio cost.
Bilingual channels
One story, one cast, released in English and a second language such as Spanish or Japanese.
The audio drama maker vs. the usual ways
| VoiceSmiths Audio Drama Studio | Generic TTS tools | Human narration | |
|---|---|---|---|
| Casting | Automatic, editable per line | Pick a voice for every line by hand | Schedule several actors |
| Emotion | Directed line by line | One tone per block | Best |
| Time for a 5,000-character chapter | About 3 minutes | An hour or more of manual work | Days to weeks |
| Cost | About $0.70–0.93 per 1,000 characters at standard prices ($0.28–0.37 in a new user's first month), by plan | Subscription plus your time | $100–300 per finished hour |
| AI label | Built in (metadata + optional spoken notice) | Up to you | Not needed |
Frequently asked questions
- What does the Audio Drama Studio do?
- It turns a chapter, a story-time script or a play into a full-cast audio drama. Narration and dialogue are separated, each line is attributed to its speaker, every character gets a distinct voice, and lines get an emotion tag where the text calls for one. Your text is never rewritten. With credits in your account you find the cast, change any speaker or emotion and preview the opening, none of which uses credits; then you render the chapter by length and export MP3 plus SRT.
- How accurate is the casting?
- Dialogue in quotes with a "she said" nearby is attributed reliably. Long runs of unattributed lines are inferred from forms of address and turn-taking, and may need a line or two corrected by hand. Every piece in the Theater was generated straight from the text, so you can judge for yourself.
- How do characters keep the same voice across chapters?
- Put every chapter in the same series: characters with the same name (and the narrator) keep their voice in every chapter, and each render remembers that chapter's cast. Matching goes by name, so write a character the same way in every chapter (always "Kaelith", not "Kaelith" in one and "Kaelith Dorne" in the next).
- Can I import a whole novel?
- Yes. In the audio drama studio, choose Import a whole novel and pick a TXT or EPUB file. It is read in your browser and split at its chapter headings; nothing is uploaded, and TXT files saved as UTF-8 or GBK both work. Each chapter shows its length and estimated credits: load one, or several short ones together (up to 8,000 characters per analysis; longer chapters are split into parts for you). When a chapter is done, Next chapter loads the following one, and a series keeps the same cast. Signed in, you can also pick a first and last chapter and add up to 50 to the render queue: the server renders them one after another with the series' voices, keeps going after you close the page, and emails you when it's done. Each chapter is charged when it starts and refunded if it fails. DRM-protected EPUBs can't be read; for a PDF, copy the text out first.
- How many voices are there? Will characters share one?
- There are 15 native Mandarin character voices and 20+ English ones. You can also design a new voice from one sentence (20 credits) or add up to 10 of the 5,000+ community voices in the workbench to your own library and cast them. When a story has more characters than voices, minor characters may share one; you can change any of them in the cast list.
- Are sound cues and scene headings read aloud?
- Today they are: lines like "SFX: door slams", scene headings and "THE END" are read as narration, so delete them before pasting; the studio doesn't add sound effects or music. Put a delivery note between the name and the colon (up to 12 characters), e.g. "Tom (whispering):". See the audio drama script format page.
- Do I need to label AI-generated audio?
- Often, yes: many platforms, including TikTok and YouTube, ask creators to disclose realistic AI-generated content. Every file the studio produces carries an embedded AI-generated label and, by default, ends with a spoken notice. You can turn the notice off before rendering, but you remain responsible for labelling wherever you publish.
AI audio drama theater
Every episode is an AI audio drama VoiceSmiths generated straight from the text. Open one and follow the script.
The Cold Palace Wine I Won't Drink
VoiceSmiths original
Deposed Jiang Wan is sent 'imperial' poisoned wine. She exposes her rival's fake pregnancy, the Emperor walks in, and the rival's last words point at him.
The Sweeper Who Shook the Sect
VoiceSmiths original
Mocked as trash for three years, Lin Yan floors the sect's top disciple with one leaf. Then an immortal reveals his clan's killer sits in the head seat.
Ding! Fired Today, System Activated
VoiceSmiths original
Framed and fired in front of the whole office, Lin Zhou hears a 'Ding!' His new System helps him strike back. But he isn't the only host in the room.