Sentioscope

Speech and voice controls, as each vendor documents them

Luma Ray: a video reference with no sound in it

The video reference names ray-2 and ray-flash-2 and lists 540p, 720p, 1080 and 4k. Sound is not in it. A separate page written for AI assistants says audio models are planned for the ecosystem, which is a roadmap rather than a control. As of 2026-09-22.

What the video reference settles, and what it never raisesThe reference names ray-2 and ray-flash-2 and lists four output sizes. Sound is absent from it, and the only sentence on the subject anywhere is a plan to integrate audio models into the ecosystem.On the referenceNot on the referenceModelsray-2 and ray-flash-2 are namedNo audio model is namedOutput540p, 720p, 1080 and 4k are listedNo track accompanies any of themSoundNever mentionedA separate page calls it a planVoicesNever mentionedNothing to attach to a characterEvery field in this register reaches this entry empty
Fig. 1 A roadmap sentence cannot be called, carries no limits and has no date, so it is logged as neither a capability nor nothing.
Luma Ray on audio source, statement by statement. Read from the vendor's video generation reference on 2026-09-22.
What the documentation settlesWhat it leaves to a take
The video reference names ray-2 and ray-flash-2 and lists four output sizesWhether any of those sizes returns a file with an audio track
Sound is not mentioned anywhere in the video referenceWhether the model produces audio that the reference simply omits
A page for AI assistants says audio models are planned for the ecosystemWhen a plan becomes a parameter, and what it will be called

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.

1A silent reference is a finding, not an omission to be filled

This entry is kept precisely because every field is empty. Dropping it would leave a table that rewards the vendors who write the most documentation, and the shape of this row carries information: the page a developer reads before committing describes picture and nothing else.

The register does not infer sound from the absence of a statement about silence, and it does not infer silence from the absence of a statement about sound. It records that the reference is about pictures, which is what can be read from it.

2A roadmap sentence is the hardest kind of claim to log

Naming audio models as a coming integration is more than nothing and much less than a documented control. It cannot be called, it has no limits, and it carries no date. Logged as a capability it would mislead; ignored entirely it would hide the one sentence on the subject the vendor has published.

So it sits in the register as what it is. A production planning dialogue work this quarter gets nothing from it. A production choosing a platform for two years might reasonably weigh it, knowing that plans move.

3The other entry that publishes nothing about sound

One more entry reaches this field with nothing on it, and for a different reason: the page that was read is about voices rather than about shots.

  • Runway — not documented by the vendor.
  • 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
  • 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

4Sources

Read from the video generation reference at docs.lumalabs.ai on 2026-09-22. The same column across every entry is on audio source; everything this vendor publishes about speech is on Luma Ray. What counts as documented is on how read.