What D-ID documents about speech

D-ID (docs.d-id.com) takes either a text script of up to forty thousand characters or an audio url, names five speech providers a voice can come from, and limits supplied audio to five minutes for a clip and ten for a talk. As of 2026-09-22.

Inclusion rule. Fields are the same five for every model here; a field the vendor does not document is recorded as not documented rather than inferred from generated output. Order. Fixed field order, identical on every model page.

1The voice is bought in, and the reference says from whom

Microsoft, ElevenLabs, Amazon, Google and Azure OpenAI are named as the providers a voice may come from. Naming them is more informative than a house catalogue would be, because a production can look the voice up at its source and know what it is getting before committing a character to it.

It also makes a dependency visible. A voice belongs to a provider, and a provider can revise its catalogue, so a series that pinned a lead to one id has a supply chain behind that character rather than a setting.

2Text or a recording, with the ceilings written down

A script can run to forty thousand characters, ten thousand of them outside markup, or a recording arrives by url. Supplied audio is limited to five minutes for a clip and ten for a talk. Those are generous next to the single-figure ceilings elsewhere in this register, and they describe a different product: a presenter delivering a script rather than a shot inside a scene.

For drama the useful reading is the shape of the limit. Long-form speech is expected here, so the timing problem that dominates generated shots barely arises, and framing takes its place as the thing to get right.

3What the reference leaves to some other page

A language field sits beside the voice, with a formal name as its example value, but the reference read for this entry carries neither a list nor a count of languages, and says nothing about what drives the mouth. Both may be documented elsewhere on the platform; this register records the page it read and dates it.

Recording the gap instead of hunting for a friendlier page is deliberate. The question a production asks is whether the thing it needs is where it would look, and the answer here is that the voice is and the language reach is not.

4This entry, one column at a time

Each of these stays inside a single field: what D-ID puts there, what the wording settles, and what it leaves for a take to answer.

• Audio source — from a script, or a supplied url.
• Voice source — a voice id, or a recording by url.
• Per-character binding — on the request.
• Languages — a language field, no list.
• Lip-sync — not documented by the vendor.

5Read against another entry

Each of these puts D-ID beside one other entry on a column where the two land at opposite grades of answer.

• Synthesia and D-ID — on lip-sync.
• D-ID and Hedra — on audio source.
• PixVerse and D-ID — on per-character binding.

6Where this model sits against the rest

Inclusion rule. Models whose vendor documents something about generated speech. A model with nothing published on the point is still listed, with the cell marked hollow. Order. Alphabetical by model name.

Inclusion rule. Models whose vendor documents something about generated speech. A model with nothing published on the point is still listed, with the cell marked hollow. Order. Alphabetical by model name.

Inclusion rule. Models whose vendor documents something about generated speech. A model with nothing published on the point is still listed, with the cell marked hollow. Order. Alphabetical by model name.

7Sources

Taken from the create a talk reference at docs.d-id.com on 2026-09-22. Every entry against the same five fields sits on the speech controls page; the rule behind each field is on how read. The entries either side of this one: Hedra, HeyGen.

• Audio source
  A script is either text of up to forty thousand characters or an audio url, with audio limited to five minutes for clips and ten for talksa script read aloud or a fileD-ID, create a talk reference / recorded 2026-09-22
• Voice source
  Five speech providers are named for the voice: Microsoft, ElevenLabs, Amazon, Google and Azure OpenAIoutside catalogues named on the pageD-ID, create a talk reference / recorded 2026-09-22
• Per-character binding
  The voice is a voice id selected from the list of available voices, with an optional language field beside ita voice id on every requestD-ID, create a talk reference / recorded 2026-09-22