What Wan 3.0 documents about speech

Wan 3.0 (alibabacloud.com) returns audio unless a caller switches it off, states that the switch does not change the price, caps reference audio at fifteen seconds, and puts the spoken line inside the prompt after the word saying. From 1 source page, read 2026-09-22; no maker pays to be listed.

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.

1A default with a price attached to it

The audio flag defaults to true, so the ordinary return from this model is a file with sound in it. The reference then does something almost nobody does: it says that enabling or disabling audio does not affect pricing. Silence costs what speech costs.

That turns a toggle into a production decision. A team rendering mute picture for a composer to score is paying for a soundtrack it discards, and knowing that in advance is worth more than a paragraph of capability copy.

2Fifteen seconds, in writing

Reference audio is WAV or MP3, one to fifteen seconds a clip, fifteen seconds in total, no larger than fifteen megabytes. The total is the number that binds: a scene with three speakers splits one fifteen-second budget between them, so voices tend to be supplied a shot at a time.

Because the clip travels on the call rather than being stored, the clip becomes a production asset in its own right. Lose it and a character's voice is gone, and nothing in a returned file records which sample produced it.

3The line goes in beside the camera direction

The worked example puts the spoken words in the prompt itself, introduced by saying, in the same breath as the description of what is in frame. Dialogue is not a field to be filled but a clause in the paragraph that also carries the blocking.

That syntax has to be written for, and it does not travel: another entry here wants the line in a block of its own, and a third takes a recording instead of words. The language of the performance is not addressed anywhere on the reference, which leaves the commonest question about a spoken line unanswered.

4This entry, one column at a time

Each of these stays inside a single field: what Wan 3.0 puts there, what the wording settles, and what it leaves for a take to answer.

• Audio source — with the picture, unless switched off.
• Voice source — reference audio, 15 seconds in total.
• Per-character binding — on the call.
• Languages — not documented by the vendor.
• Lip-sync — not documented by the vendor.

5Read against another entry

Each of these puts Wan 3.0 beside one other entry on a column where the two land at opposite grades of answer.

• Wan 3.0 and Wan2.2-S2V — on audio source.
• MiniMax and Wan 3.0 — on voice source.
• Sora 2 and Wan 3.0 — on audio source.

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 video generation API reference at alibabacloud.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: Wan2.2-S2V, D-ID.

• Audio source
  The audio parameter defaults to true so the returned video carries an audio track, and false returns one withouton unless it is switched offAlibaba Cloud, Wan 3.0 API reference / recorded 2026-09-22
• What choosing silence costs
  Enabling or disabling audio does not affect pricingAlibaba Cloud, Wan 3.0 API reference / recorded 2026-09-22
• Voice source
  Reference audio is WAV or MP3, one to fifteen seconds a clip, fifteen seconds in total and no more than 15 MBa published capAlibaba Cloud, Wan 3.0 API reference / recorded 2026-09-22
• How a spoken line is written into a prompt
  The worked example puts the spoken line in the prompt itself, after the word sayingAlibaba Cloud, Wan 3.0 API reference / recorded 2026-09-22