use case
verified July 2026
Voice the whole cast, re-read every line note.
Animation is an ensemble problem. A cast is many distinct voices, a cut is many takes of each, and both numbers climb with every revision pass. A meter prices the work by the two things a production cannot hold down — the size of its cast and the number of times a line gets a note.
01
The arithmetic of an ensemble
A short film or a motion comic is a handful of characters; a full animation is a cast, and each character needs a distinct voice held steady across every scene. Instant voice cloning turns each one into a stored reference, so the cast is a set of parameters a single seat carries rather than a schedule of booked sessions.
The revisions are where a meter bites. Animation is cut and re-cut, and a line note means re-reading the same line with a different emphasis, a different beat, sometimes a different emotion. On an unmetered seat the re-read costs the same as the first take, so a note becomes a normal edit rather than a budget conversation, and concurrent seats let the ensemble be built in parallel.
02
The controls a directed read needs
- Each character is a cloned reference, so switching who is speaking is switching one field, and a series voice does not drift between sessions.
- Sixteen emotion names over eight tuned presets, so one character can carry a scene arc — tired, then frightened, then resolved — without a second voice.
- temperature sets melodic range and cfg_weight loosens the pacing, so a line note about delivery is a parameter change rather than a re-record.
- Reproducible seeds hold a delivered line steady while the lines around it are re-cut, and break tags place the beat exactly where the animatic wants it.
03
Notes — an engineer's checklist
01How do I voice several characters from one seat?
Each voice is a request parameter conditioned on its own reference, so a single seat holds the whole cast and moving between characters is moving one field rather than opening a new session.
02What happens when a director sends a line note?
A note is a re-render. You change the emphasis, the emotion or the prosody and run the request again, and the new take costs the same as the first, so revisions are not rationed.
03Can a character stay consistent across a whole series?
The voice is a stored reference, so a character conditioned on it does not drift between episodes or across a long production, and a reproducible seed keeps a signed-off line unchanged.
See also — related sheets
capability
0
Voice cloning from ten seconds, in the request
Zero-shot voice cloning: a ten-second reference rides inside each request, no training job, and the identity holds across 23 languages.
capability
7.9
Delivery is a request parameter
The prosody dial moves measured pitch range from 10.1 to 18.0 semitones on one voice, one sentence — continuous per-request control on the production API.
who it’s for
1
Gandr for localisation and dubbing studios
Carry one speaker across 23 languages and price a back catalogue by the seats it runs on rather than by its total length.
use case
16
Every line voiced, including the ones nobody planned
Game dialogue on a flat seat: voice the whole script including barks and branch lines, and re-render after a writing pass without paying per character.
A key and one seat to build this on — the same production API this page measures.
Request access for this use case