Hades
English · Narration
For game masters prepping a session
Write the villain's threat, the oracle's riddle and the innkeeper's rumour. FullCastBook gives every NPC a different AI voice and a narrator the read-aloud text, so you press play when the moment comes.

Lewis Carroll
Alice's Adventures in Wonderland
Alice asks the grinning Cheshire Cat for directions — and gets none.
The Cat only grinned when it saw Alice. It looked good-natured, she thought; still it had very long claws and a great many teeth, so she felt it ought to be treated with respect.
Step-by-step guide
No card to sign up, nothing to install.
Hear an example
Generated from the script below with the studio's own voices and settings, one voice per speaker, joined the way an export joins lines. No edits afterwards.
Narrator · Vessa · Bram · Sister Maelis · Voices: Gemini 3.8 Flash TTS
D&D NPC Voices
What you get, from manuscript to finished audiobook.
Script template
Replace the text in brackets with your own, paste it into the studio, and every speaker gets a voice.
One Part per clip you want to play. NPC name in capitals before a colon; boxed text goes to the NARRATOR; delivery notes in parentheses or brackets.
Part 1: [LOCATION] arrival NARRATOR: (read-aloud) [WHAT THE PARTY SEES, HEARS AND SMELLS, IN TWO OR THREE SHORT SENTENCES] [NPC NAME]: (gruff, wiping a tankard) [GREETING] [NPC NAME]: [THE RUMOUR OR HOOK THEY OFFER] Part 2: [VILLAIN] message NARRATOR: [HOW THE MESSAGE ARRIVES: a sending stone glows, a mirror clouds over] [VILLAIN]: (calm, amused) [GREETING THAT USES A PLAYER CHARACTER'S NAME] [VILLAIN]: [ONE SPECIFIC THREAT OR OFFER] [VILLAIN]: [laugh] [DETAIL THAT SHOWS THEY HAVE BEEN WATCHING] [VILLAIN]: [whispers] [PARTING LINE] Part 3: [ORACLE OR ARTIFACT] prophecy [ORACLE]: (distant, echoing) [PROPHECY LINE 1] [ORACLE]: [PROPHECY LINE 2] [ORACLE]: [sigh] [THE LINE THEY WILL ARGUE ABOUT FOR WEEKS] Part 4: [VILLAIN] alternate, if the party refuses [VILLAIN]: (cold, no longer amused) [WHAT HAPPENS NOW]
Each Part line starts a new chapter and its title is read at the top of that file; trim it, or name the Part something you're happy to hear.
Use cases
Where authors and storytellers put it to work.

In practice


Demo performances. In the studio you cast from standard and premium voice lists.
English · Narration
English · Narration
English · Character
English · Character
English · Narration
English · Narration
English · Narration
English · Narration
English · Narration
English · Narration
In depth
D&D NPC voices are one of the hardest parts of running a game. You are already tracking initiative, rules and the plot, and then the party meets their fourth shopkeeper of the night. Most game masters keep a note per NPC (deep voice, slow, never smiles) and improvise. For a few moments a session, though, a prepared recording can do what improvising can't: a villain who sounds nothing like you, a prophecy delivered exactly as written, a message that arrives with real menace.
FullCastBook is built for that prep. Write the lines, and the studio gives every NPC a different AI voice and a narrator the read-aloud text, then exports clips you play at the table. It is the same engine as multi-voice text to speech, set up for game masters. It is not a live voice changer: everything is generated before the session.
Record the speeches players can't interrupt. Messages arriving through a sending stone, a letter read aloud, a vision, a ghost replaying its last words, the villain's speech from a balcony: these are one-way, so a recording loses nothing. The same goes for set-piece boxed text, the description you rewrote three times to get the mood right.
Keep conversations live. An NPC the party will question, haggle with or try to deceive needs your improvisation, and players notice when a recorded voice gives way to yours mid-scene. If an NPC will speak both ways, choose an AI voice close to how you'd play them yourself, or keep the recording to their entrance and take over from there. Session zero is a good time to ask whether your table enjoys recorded moments at all; some groups love them, others prefer everything live.
Contrast is what makes NPCs memorable. Give the gravelly old captain, the bright young acolyte and the smooth-talking noble voices far apart in pitch and pace, and keep the narrator calmer than all of them. If two recurring NPCs appear together, test their first lines back to back before you commit. The one-line voice note you already keep for each NPC makes a good bracket direction.
Fantasy names trip any voice, so spell them the way they sound: Vessa, not Vhessah; Kael-thas with a hyphen if that's how you say it. The studio voices what you write, so the script can use phonetic spellings while your notes keep the canon one. For more on building voices by archetype, see how to do character voices.
A good villain monologue is short. Thirty seconds of a calm voice that knows the party's names is worse to hear than three minutes of ranting. Write it for the ear: short sentences, one specific threat, one detail that shows the villain has been watching, and a parting line. Then direct it: [silky, in no hurry] on the greeting, [laugh] before the cruel line, [whispers] for the last word.
Boxed text works the same way. Read the published version aloud once and you will hear where it runs long. Trim it, split it into a few sentences per line and give it to the narrator. If an NPC speaks inside the scene, put their name before their line and they get their own voice. More direction ideas are in AI voice emotion tags.
Exports come as audio per chapter, not per line, so plan your clips as chapters. Start each one with a heading line such as Part 3: the oracle, and the ZIP export gives you one numbered MP3 per clip. The heading is read at the top of each file, so give it a name you don't mind hearing or trim the first second in an editor.
At an in-person table, play the files from your phone or laptop through a speaker. Online, load them into your virtual tabletop's playlist, a soundboard app or a music bot on Discord, so every player hears the same thing at the same time. Give each Part a name you can spot mid-fight: 03 - oracle is easier to find under pressure than track 3, and the numbering keeps them in session order. There is no music or ambience in the export; layer that from your usual source.
Keep the whole campaign in one project and recurring NPCs keep their voices for months. The same project can open each session with a spoken "previously on", covered in D&D session recap audio.
Your homebrew NPCs and your own lines are yours. Read-aloud text from a published adventure belongs to its publisher: voicing it for your own table is one thing, posting it in a public actual-play episode is closer to copying, so check the publisher's fan content policy first. Generating and exporting use credits; see pricing for plans and top-ups.
Tips
Yes, for lines you prepare in advance. Write the NPC lines in FullCastBook, give each NPC and villain a different AI voice and export clips to play at the table. It doesn't change your voice live.
No, there are no real-time voices. You generate clips before the session and play them when the moment comes, which suits villain messages, visions and boxed text best.
Export a ZIP with one MP3 per Part, then add the files to your virtual tabletop's playlist, a soundboard app or a music bot on your Discord server. Test the volume before the session.
Pick a calm, controlled voice rather than a loud one, keep the speech short, and direct it with notes like [silky, in no hurry] and tags like [laugh] or [whispers] on the key lines.
Read-aloud text belongs to the adventure's publisher, so your own words are the safe choice. Playing a clip privately at your table is very different from posting it publicly, for example in an actual-play podcast; check the publisher's fan content policy first. This is general information, not legal advice.
Yes. Keep the campaign in one project and every recurring NPC keeps their voice, session after session. The same project can hold your session recaps.
They can, for a recap or a trailer, but ask your players first. Most players want to voice their own characters at the table, so keep AI voices for NPCs during play.
No. Signing up is free with no card and voice samples are free to hear, but generating and exporting audio use credits. See pricing.
Write the lines, cast your NPCs and export clips for the table. Free to sign up; pay with credits when you create.