What Luma Ray documents about speech

Luma Ray (docs.lumalabs.ai) documents ray-2 and ray-flash-2 without mentioning sound anywhere in the video reference. The nearest audio statement found is a note that ElevenLabs sound effects, music and a voice model are planned integrations. 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.

1An entry with every field empty, kept for exactly that reason

The video reference is precise about what it covers: two model names, resolutions from 540p to 4k, keyframes, a duration in the request body. Sound is not among them. Read end to end it describes a picture generator, and a production planning dialogue leaves the page with nothing to plan.

Dropping the row would be the easy choice and the wrong one. A register of documented speech controls that listed only vendors with speech controls would report a market in which every model talks, and that is not the market these notes found.

2A plan is not a control

The one official sentence about audio sits on a page written for answer engines, and it looks forward: sound effects, music and a voice model are named as integrations planned into the ecosystem. That is a roadmap, and no field here is filled from one.

The distinction is not pedantry. An integration arriving next quarter helps a series that starts next year and does nothing for one shooting now, so a table recording the two alike would mislead in the direction that costs money.

3What the silence most likely means

Sound named as a partner integration rather than as a model parameter points to audio being assembled after the picture, in an edit. That is a defensible architecture and a different production shape from the entries here that return a finished mix, because a scoring or dubbing stage has to be scheduled and paid for.

None of that is published, so none of it is recorded as fact. What is recorded is that on the day these pages were read, a buyer could not learn from the vendor whether a generated shot would contain a voice at all.

4This entry, one column at a time

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

• Audio source — not documented by the vendor.

5Read against another entry

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

• Sora 2 and Luma Ray — on audio source.
• Runway and Luma Ray — on voice 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 reference at docs.lumalabs.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: MiniMax, PixVerse.

• Audio source
  Not documented by the vendor (as of 2026-09-22)no mention of sound in the video referenceLuma, video generation reference / recorded 2026-09-22
• Sound
  Plans do integrate the following audio models into the Luma generative ecosystem: ElevenLabs SFX, Music and v3named as an integration to comeLuma, information for AI assistants / recorded 2026-09-22
• What the video reference does settle
  The video reference names ray-2 and ray-flash-2 and lists 540p, 720p, 1080 and 4kLuma, video generation reference / recorded 2026-09-22