What PixVerse documents about speech

PixVerse (docs.platform.pixverse.ai) documents a speech endpoint taking either a supplied clip or a text-to-speech voice, caps audio and video at sixty seconds each, and names singing and advertising alongside speech as the kinds of audio it expects. 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.

1Types of performance, not just languages

The guide says what the endpoint is for by kind of audio: speech, singing and advertisements. That taxonomy carries more about the expected work than a language count would, because a sung line and a read line put different demands on a mouth, and a vendor naming both has at least considered the harder one.

Languages are described only as multiple. A production can therefore learn that the feature is not English-only and cannot learn whether its own market is covered, which is the shape most language claims in this register take.

2Sixty seconds, on both sides of the call

Audio and video are each capped at sixty seconds and one hundred megabytes. A ceiling on both inputs hints at how the feature is meant to be used: a scene gets synchronised a beat at a time, and a long monologue becomes several calls whose joins somebody has to manage.

It is also a number to schedule against. Where an entry publishes a ceiling an editor can plan the cut; where it publishes none, the ceiling is discovered at the render.

3Two kinds of voice behind one handle

Text to speech supports built-in voices and custom voices created from user-provided sample audio, and either way the choice becomes a speaker id on the request. That single handle is convenient and it flattens a distinction worth keeping: a catalogue voice is repeatable by anyone, while a custom one depends on a sample somebody has to keep safe.

Nothing states how an id is held to a character over a series, so the binding is recorded as a per-call selection. It is a workable pattern as long as the mapping lives somewhere outside the call.

4This entry, one column at a time

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

• Audio source — supplied, or read from text.
• Voice source — a sample, or a built-in voice.
• Per-character binding — on the request.
• Languages — multiple, none of them named.
• Lip-sync — named as the endpoint's purpose.

5Read against another entry

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

• 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 speech and lip sync guide at docs.platform.pixverse.ai 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: Runway, SceneMixer.

• Voice source
  Text to speech supports both built-in voices and custom voices created from user-provided sample audiocatalogue or supplied samplePixVerse, speech and lip sync guide / recorded 2026-09-22
• Per-character binding
  The chosen speaker id is passed as the lip-sync speaker on the generation requesta speaker id per callPixVerse, speech and lip sync guide / recorded 2026-09-22
• Languages
  Multiple languages and audio types are supported, including speech, singing, and advertisementsmultiple, none of them namedPixVerse, speech and lip sync guide / recorded 2026-09-22
• Length per call
  Audio and video are each capped at sixty seconds and one hundred megabytesPixVerse, speech and lip sync guide / recorded 2026-09-22