Skip to main content

VEED Fabric 1.0 turns one still image and one speech track into a lip-synced talking video at 480p or 720p, for photos, paintings and mascots alike. Billed per second of audio.

From $0.096/s

Workbench

Input

Talking avatar
1/1

Drop files or media links, paste, or browse · 1 file · Max 10 MB · JPG · PNG · WEBP

  • 1
1 / 1 files
Required. JPEG, PNG or WebP with one clearly visible face. The video takes the aspect ratio of this image. at least 1 characters
1/1
1 / 1 files
Required. Speech in MP3, WAV or another common format, up to five minutes. Its length sets the length of the clip and the charge. at least 1 characters
affects price
480p or 720p. 720p costs more per second; both keep the aspect ratio of the image.
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
Example result

Related models and next tasks

Missing the model you need?

IMAGE + VOICE TO VIDEO

One still image and one voice track become a talking, lip-synced clip.

Key numbers

Inputs
2
One image and one speech file. There is no prompt to write.
Tiers
720p
480p for drafts, 720p for the final cut; the frame keeps the shape of your image.
Audio
5 min
The clip lasts exactly as long as the voice track you send.

Before you spend

Talking avatar

One image and one audio file per request, with audio up to five minutes. The face should be clearly visible; profiles, tiny faces and covered mouths animate poorly. There is no prompt, no camera control and no background replacement. Text in the audio is not read from a script, so send recorded or synthesised speech.

What it does

A painting that talks

An 18th-century portrait speaks a ten-second line

We gave Fabric a 3:4 oil portrait and a recorded female voice. The lips, the jaw and the tilt of the head follow the words, while the brushstrokes, the craquelure and the emerald silk stay as painted. The 720p render came back at 832×1088.

The input

The still we started from

The input image for the VEED Fabric 1.0 example: an 18th-century style oil portrait of a young woman with auburn curls in an off-the-shoulder emerald silk gown, holding a closed lace fan.

This is the 3:4 portrait we sent as image_url. Fabric kept its colours, its framing and its paint texture, and added only the speaking motion on top.

No rig, no prompt

The voice track does the directing

Pauses in the recording become pauses on screen and a rising sentence lifts the head a little. Change the audio and the same image gives a different performance.

Where it fits

  • Characters that greet visitors

    Give a brand mascot, a game character or a virtual host a voice and a moving face without building a 3D rig.

  • Old pictures that talk

    A painting, a vintage photo or an illustrated poster can deliver a narrated line for a museum piece, an ad or a social post.

  • Personal video messages at scale

    Swap the voice track to produce one clip per customer, language or offer from the same portrait.

Model reference

VEED Fabric 1.0 API reference

VEED Fabric 1.0 API specifications

Model
VEED Fabric 1.0
Required inputs
Image (image_url) and speech audio (audio_url)
Resolution
480p or 720p, aspect ratio taken from the image
Length
Same as the audio, up to 5 minutes
Output
MP4, H.264 with AAC audio, 25 fps
Billing
Per second of audio, rounded up

How to call the VEED Fabric 1.0 API

  1. Pick a picture with a readable face

    A face that looks towards the camera and fills a fair part of the frame animates best. Photos, oil portraits, anime stills and mascots all work.

  2. Record or generate the line

    Fabric speaks audio, not text. Record the voice or produce it with a text-to-speech model, then pass the file link in audio_url.

  3. Choose the resolution

    Start at 480p while you test timing, then render the keeper at 720p. Both tiers follow the aspect ratio of the image.

VEED Fabric 1.0 API questions