GET

/v1/schemas/fal/video/fal-ai/sync-lipsync/v3

Accept:

kind inputactivity videohash 317f3ef4…extracted Oct 22, 2026version currentsource upstream spec ↗

field	recorded source
schema	Source linked above. upstream-spec Source trace `15b7ee2d29220ab82833cd31606d0ac2dd580b6ecdc6fd857a7077ccd5f7b603` { "fetchedAt": 1792645225, "extractorVersion": "1", "verifiedAt": null }

Request body

5 properties · 2 required · 2 $defs

property	type	description	constraints
audio_url*	string	URL of the input audio
video_url*	string	URL of the input video
options	Sync3GenerationOptions	Additional Sync.so generation options. If options.sync_mode is set, it overrides the top-level sync_mode field.	nullable
sync_mode	enum	Lipsync mode when audio and video durations are out of sync.	cut_off \| loop \| bounce \| silence \| remap · default: "cut_off"
enable_safety_checker	boolean		default: true

Types

2 $defs

Sync3ActiveSpeakerDetectionobject

coordinates	integer[]	Reference point [x, y] on the speaker's face in frame_number.	items ≥ 2 · items ≤ 2 · nullable
auto_detect	boolean	Whether to automatically detect and apply generation to the active speaker.	nullable
bounding_boxes_url	string	URL to a JSON file containing per-frame bounding boxes.	nullable
v3	boolean	Whether to use Sync.so's active speaker detection v3.	nullable
face_image	string	Base64 encoded face image for manual speaker selection.	nullable
use_v2	boolean	Deprecated Sync.so option for active speaker detection v2.	nullable
frame_number	integer	Frame index for manual speaker selection coordinates.	n ≥ 0 · nullable
bounding_boxes	integer[][]	Per-frame bounding boxes [x1, y1, x2, y2] for the detected face.	nullable

Sync3GenerationOptionsobject

temperature	number	Controls how expressive lipsync can be; 0 is least expressive, 1 is most expressive. Sync.so handles this natively for sync-3, so this option is ignored by sync-3.	0 ≤ n ≤ 1 · nullable
active_speaker_detection	Sync3ActiveSpeakerDetection	Active speaker detection configuration for videos with multiple people.	nullable
occlusion_detection_enabled	boolean	Whether to detect occlusion during generation. Sync.so handles this automatically for sync-3, so this option is ignored by sync-3.	nullable
prompt	enum	Emotion prompt for generation. This is a Sync.so option primarily used by react-1.	happy \| sad \| angry \| disgusted \| surprised \| neutral · nullable
sync_mode	enum	Defines how to handle duration mismatches between video and audio inputs. Overrides the top-level sync_mode when provided.	cut_off \| loop \| bounce \| silence \| remap · nullable
model_mode	enum	Edit region for the model. This is a Sync.so option primarily used by react-1.	lips \| face \| head \| lipsync \| emotion \| talking_head · nullable

Validate a payload

before you spend tokens

shell

$ curl -X POST https://modelschemas.com/v1/validate \
    -d '{"provider":"fal","endpointId":"fal-ai/sync-lipsync/v3","payload":{…}}'
→ {"valid": false, "errors": [{"path": "#", "keyword": "required", …}]}