Synthesia: a voice id, an avatar, and no stated pairing

Both halves of a character exist here as documented objects. Voices carry ids in a published catalogue and avatars are created and reused. What is missing is the sentence that joins them, which is the odd gap in an otherwise structured platform. As of 2026-09-22.

Fig. 1 If framing affects how a spoken line lands on a face, pitch and pace plausibly do too, and nothing addresses whether some voices suit some avatars.

Synthesia on per-character binding, statement by statement. Read from the vendor's voices reference and avatar documentation on 2026-09-22.
What the documentation settles	What it leaves to a take
Each voice is listed with a language name, a code, a gender, a name and an id	Which id belongs to which avatar by default, if any
Lip sync and facial expressions are generated from the spoken content	Whether the avatar constrains which voices suit it
Voice cloning carries a language list of its own	Whether a cloned voice can be made the avatar's own

Inclusion rule. Only statements found on the vendor page named in the caption, and only ones bearing on this one field. Nothing here is filled in from generated output. Order. In the order the statements arrive on the vendor's page.

1Two well-documented objects and a missing join

Most blanks in this column belong to products that document almost nothing about voices. This one documents them thoroughly and still leaves the pairing unstated, which is the harder kind of gap to read: the pieces are all present and their relationship is assumed.

In practice a production will pass a voice id with an avatar on every request and treat the combination as the character. That works, and it means the character lives in the calling code rather than on the platform.

2The framing condition hints that pairing matters

Lip sync is documented as performing better with the avatar framed closer to camera, which says the relationship between a face and a spoken line has conditions on it. If framing affects it, a voice's pitch and pace plausibly do too, and nothing addresses whether some voices suit some avatars.

That is left as a question rather than inferred. The register logs that both objects are published with ids and that no document connects a given pair.

3Others whose documentation never addresses persistence

Seven more entries reach this column empty. This is the only one where both a voice and a character are published as durable objects and the link between them is not.

Hedra — not documented by the vendor.
Luma Ray — not documented by the vendor.
SceneMixer — not documented by the vendor.
sync-3 — not documented by the vendor.
Veo — not documented by the vendor.
Vidu — not documented by the vendor.
Wan2.2-S2V — not documented by the vendor.

Languages
Each voice is listed with its formal and native language name, a language code, a gender, a name and a voice idpublished as a catalogue with codesSynthesia, list of supported voices / recorded 2026-09-22
A framing condition attached to lip-sync
Lip sync performs best when the avatar is framed closer in the scene rather than positioned far from the cameraSynthesia, create an avatar / recorded 2026-09-22
Voice source
Voice cloning carries a language list of its own, running from Afrikaans to Zulu and described as the same list used for standalone cloninga cloned voice with its own reachSynthesia, create an avatar / recorded 2026-09-22

4Sources

Read from the voices reference and avatar documentation at docs.synthesia.io on 2026-09-22. The same column across every entry is on per-character binding; everything this vendor publishes about speech is on Synthesia. What counts as documented is on how read.