Sentioscope

Speech and voice controls, as each vendor documents them

Entries whose pages never mention sound at all

Two entries fill nothing. One publishes a video reference that never mentions sound; the other publishes a voices reference that never mentions a shot. Neither blank is evidence about a product, and both are evidence about a document. As of 2026-09-22.

Two empty rows, arrived at from opposite directionsOne page is about generating video, so it lists models and resolutions and stops. The other is about constructing voices, so it lists sample windows and byte ceilings and stops. Neither covers the join between them.A video referenceA voices referenceWhat it settlesModels, and four output sizesSample windows, ceilings, an id, apreviewWhat it skipsSound, entirelyAny picture at allStrongest statement on soundAudio models named as a planA preview you can listen toWhy the row is keptThe shape of a blank row is a findingThe sameBoth halves of the subject, and neither covers the middle
Fig. 1 That join is where this register lives, and a production is the one left to work it out.
The two entries with the emptiest rows, and what each page was actually written about. Recorded 2026-09-22.
ModelWhat the page that was read is aboutAnd what it says a production may supply
Luma RayNot documented by the vendorNothing documented as an input
RunwayNot documented by the vendorA 10-second to 5-minute sample, or a sentence

Inclusion rule. Entries where the vendor page read for this register addresses neither where sound comes from nor how a mouth behaves. An entry with one filled field is not listed here. Order. Alphabetical by model name.

1Keeping sparse rows is what makes the table honest

Dropping entries that publish little would leave a comparison that rewards whoever writes the most documentation. The shape of an empty row is itself a finding: it says the page a buyer reads before committing does not address the thing this register is about.

Neither of these is a weak product. One is a well-known picture generator and the other has the most developed voice-construction reference here. The blanks describe editorial priorities, and those are what a reader has to plan from.

2The two blanks come from opposite directions

One page is about generating video, so it lists models and resolutions and stops. The other is about constructing voices, so it lists sample windows and byte ceilings and stops. Between them they cover both halves of the subject and neither covers the join.

That join is where this register lives. Whether sound comes out with the picture, and whether a voice survives to the next episode, are questions that fall between a video reference and a voice reference, and a production is the one left to answer them.

3A roadmap sentence is logged as neither a capability nor nothing

One of these has a page for AI assistants naming audio models as a planned integration. It cannot be called, has no limits and carries no date, so recording it as a capability would mislead and ignoring it would hide the only published sentence on the subject.

The other has a preview step, which is a test rather than a specification. Both are worth noting precisely because they are the strongest things either vendor offers in a column otherwise empty.

4The entries on this route, one page each

Each of these links to that entry on the field this route groups by. The wording behind every cell, and the date it was read, sits on the page it links to.

  • Luma Ray — not documented by the vendor.
  • Runway — not documented by the vendor.
  • 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
  • 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
  • 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
  • Lip-sync
    Not documented by the vendor (as of 2026-09-22)not addressed where voices are definedRunway, custom voices reference / recorded 2026-09-22
  • Per-character binding
    A voice is created asynchronously, reaches a ready state with a preview, and is then referred to by its own ida stored voice objectRunway, custom voices reference / recorded 2026-09-22
  • Voice source
    A voice can instead be asked for in words, with the description required to run at least 20 charactersa route that needs no recordingRunway, custom voices reference / recorded 2026-09-22

5Sources

Membership of this route is decided by the wording in the field it groups by, read from the documentation each vendor publishes. The column itself is on the field note, all five columns are on the speech table, and what counts as documented is on how read. Other routes: A voice that is stored, A voice per call.