Runway: voices get previews, performances get nothing
This reference is unusually thorough about voices and says nothing about what happens to a face. It defines what a voice is made from and how it is addressed, and stops before the performance begins. As of 2026-09-22.
| What the documentation settles | What it leaves to a take |
|---|---|
| A voice is created asynchronously and reaches a ready state with a preview | What the preview reveals about how a line will look |
| A custom voice comes from a 10-second to 5-minute sample, or a description of 20 characters | Whether a described voice aligns as well as a sampled one |
| Nothing on the page addresses lip movement | Where in the product a voice meets a picture at all |
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 reference that ends where this column begins
The page is about constructing voices. It gives a sample window, a byte ceiling, an alternative route that needs no recording, and a lifecycle with a preview. Then it stops, because how a voice is applied to footage is somebody else's page.
That makes this blank easier to explain than most and no less real. A production comparing tools on this column has nothing to compare, and the register does not fill the gap from the existence of a voice system.
2A preview answers a different question well
Hearing a voice before any shot is made is the one documented audition step in this register, and it settles casting. What it cannot settle is whether the mouth will look right, because a preview is audio.
So this entry sits at opposite ends of two columns: the fullest voice-construction answer here, and an empty cell on the thing an audience notices first. The pairing is worth recording, because it says what the product was designed around.
3Others whose pages never raise the subject
Six more entries reach this column empty. Most of those pages are about generating pictures; this one and one other are about the sound and still leave the mouth out.
- D-ID — not documented by the vendor.
- LTX Studio — not documented by the vendor.
- Luma Ray — not documented by the vendor.
- Veo — not documented by the vendor.
- Vidu — not documented by the vendor.
- Wan 3.0 — not documented by the vendor.
- Lip-syncNot documented by the vendor (as of 2026-09-22)not addressed where voices are defined
- Per-character bindingA voice is created asynchronously, reaches a ready state with a preview, and is then referred to by its own ida stored voice object
- Voice sourceA custom voice can be built from an audio sample between 10 seconds and 5 minutes long and at most 10 MBa published sample window
4Sources
Read from the custom voices reference at docs.dev.runwayml.com on 2026-09-22. The same column across every entry is on lip-sync; everything this vendor publishes about speech is on Runway. What counts as documented is on how read.