Voice id: a handle that outlives one call
A voice id is a name for a voice that survives the request it was used on. Five entries publish one. Because an identifier cannot drift, a production that records it gets a repeatable voice without storing any audio. As of 2026-09-12.
| Model | What the id refers to | What is not published about it |
|---|---|---|
| D-ID | A voice from one of five outside providers | Whether it survives an upstream reorganisation |
| HeyGen | An enrolled clone | Whether a clone can be revised without a new id |
| PixVerse | A built-in voice, or one from your sample | Whether re-pointing it changes the voice |
| Runway | A voice created asynchronously, with a preview | Whether the same description recreates it |
Inclusion rule. Entries whose documentation refers to a voice by an identifier passed on a request. A reference clip supplied per call is not an identifier and does not earn a row. Order. Alphabetical by model name.
1Why an identifier beats a recording for continuity
A clip supplied on every call has to be the same file every time, and nothing checks that it is. An identifier is a string: pass the same one and the same voice arrives, so the continuity problem becomes a spreadsheet problem rather than a listening problem.
The catch is that nothing in a returned file records which id was used. Passing the wrong one produces a valid call and a different person, so a shot list remains the authoritative record of a cast whichever mechanism a platform offers.
2Nobody publishes a lifetime for one
Across all five entries, no document says whether an id survives a catalogue change, a plan change or a model version. That is the single most useful sentence none of them has written, because a series is a long time to depend on an unpromised string.
The defensive habit is to treat every id as immutable. Never re-point one at a new sample; create another. That keeps a production's own record honest even where the platform makes no commitment.
3An id is a voice object, not a character
Where the identifier belongs to a voice, the pairing between a person in the story and that voice lives in a production's records. Where a vendor binds the voice to a character object instead, there is nothing to pass and nothing to pass wrongly.
The first is more flexible and more clerical: a voice can be reused across series, shared deliberately, or retired without touching the character. Somebody has to keep the list either way.
4Voices in a drama pipeline
For the speech side of “Voice id”, SceneMixer's guides describe how lines are spoken:
It is a property of the character, written once in the voice description, and it covers every line they speak. What changes line by line — mood, pace, a word under stress — goes in the delivery note in brackets at the start of that line.
SceneMixer, Give a Character an Accent or Dialect in an AI Drama — “Do I have to repeat the accent on every line?” (read 2026-10-13)
Each character is matched to a preset voice sample that is attached to every segment, so the character's voice identity is stable across episodes.
SceneMixer, Native-Language Dialogue vs. Dubbing in AI Video (2026): Why the Cast Should Speak the Line — “Does the character keep the same voice across languages?” (read 2026-10-13)
5Sources read for this entry
This page defines the term and logs what each vendor documents about it, each figure read on 2026-09-12. Related: Voice cloning, Text to speech.