Does Kling motion control make a photo talk? Where sound comes from

Kling 3.0 motion control does not turn a photo into a speaker from a script or an audio file: it copies the movement of a reference video onto your still. Sume's route takes image_url or an avatar, a motion_video_url that "drives all output motion", duration_seconds and an optional appearance prompt, and no audio_url. For a face that speaks given audio, use VEED Fabric 1.0 or MiniMax H3 Max Lip Sync. A 12-second motion clip costs $1.89 on Sume.

What the sound flag does

keep_original_sound defaults to true, which keeps the sound of the reference clip in the output; false gives a silent clip. That is the reference clip's audio, not speech generated for your photo. The fal page lists the model as transferring movement from a reference video to a character image, at $0.126 per second of output; Sume bills ceil(duration_seconds) x that list x 1.25, or $0.1575 per second.

Pick the route by input

The table matches what you have to the Sume route.

Motion-control request

The reference video sets the output length; duration_seconds is the pricing basis from 1 to 30.

curl -X POST https://api.sume.com/v1/kling/3.0/motion-control \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: kling-mc-001" \
  -d '{
    "image_url": "https://example.com/presenter.png",
    "motion_video_url": "https://example.com/gesture.mp4",
    "duration_seconds": 12,
    "keep_original_sound": false
  }'