Introduction
Pair existing face footage with replacement MP3 or WAV speech to create a revised MP4. Choose among four timing modes; the complete uploaded audio duration determines billing even when cut_off shortens the result.
| Field | Value |
|---|---|
| Input | Video, Audio |
| Output | Video |
| Publisher / tier | Sync / Lipsync 2 Pro |
| Request model | sync/lipsync-2-pro/lip-sync |
| Required media | Face video and replacement speech |
| Recording bounds | 1–300 seconds; MP3 or WAV |
| Optional mode |
|
| Billable quantity | Uploaded audio duration rounded up |
| Delivery | Asynchronous task; MP4 output |
| Example dimensions | 1280×720 at 24 fps |
| Review refusals | No charge |
| Output availability | 14 days before deletion |
Key features & highlights
Watch the watchmaker deliver a new line
In our tests, a 6.04-second watchmaker shot paired with 4.2 seconds of speech produced a 4.21-second result. Lip detail, teeth, and salt-and-pepper stubble remained visible as his mouth followed the replacement words. The MP4 measured 1280×720 at 24 fps.
Assess detail where the face fills the frame
Sync says Lipsync 2 Pro emphasizes fine facial detail while retaining the speaker's delivery style without speaker-specific training. A close-up gives an editor a place to inspect that claim in motion. The accompanying cover is an illustration, not a test output.
A trimmed result can still bill the full recording
cut_off stops the result at the shorter input, but billing counts the audio you submitted, rounded up. Trim an overlong recording before submission, or use loop, bounce, or remap when the picture needs to span the line. Changing the mode does not change the rate.
Request parameters
| Parameter | Required | Description |
|---|---|---|
video_url | Yes | Required URL for the footage; no default. Choose a clearly visible face, especially for a close-up detail check. at least 1 characters |
audio_url | Yes | Required MP3 or WAV URL with 1–300 seconds of replacement speech; no default. Applies with every |
sync_mode | No | Optional alignment choice: |
Pricing
$0.101 / second
Charges count the uploaded audio duration, rounded up to whole seconds, for recordings from 1 to 300 seconds. Video length and sync_mode do not affect that calculation; cut_off can shorten the output while the full audio remains billable. Content-review refusals incur no charge.
Use cases
Revise a watchmaker scene
Use the supplied close-up as a reference for a short dialogue edit. Inspect the moving mouth, teeth, and stubble while checking that the replacement line reaches its intended ending.
Replace an ad read in close-up
Record revised wording for footage featuring an authorized spokesperson. Pro targets the fine facial detail visible in a tight shot; inspect the delivered edit before using it in the ad.
Dub a filmed conversation excerpt
Prepare the replacement speech recording for the selected shot. Sync says Pro retains the speaking style visible in the source; choose a timing mode that fits the new recording.
Fit dialogue to an available take
Use cut_off for a shorter ending, loop for repeated picture, bounce for alternating playback, or remap to adjust the video timeline. Review any repetition or speed change against the intended scene.
Prompt guide & pro tips
Supply the two media URLs
Set
video_urlto clear face footage andaudio_urlto an MP3 or WAV recording lasting 1–300 seconds. Measure the recording because it sets the billable duration.Choose a mode and submit
Select one of the four supported
sync_modevalues, or omit it forcut_off. Submit the task with model sync/lipsync-2-pro/lip-sync and retain its identifier.Retrieve and examine the result
Wait for completion through polling or a callback. Check mouth alignment, fine detail, and the ending, then save the MP4 before its 14-day retention expires.
Close-up dialogue editing with Pro
Sync Lipsync 2 Pro is Sync's detail-focused tier for changing the visible speech in existing footage. Sync describes it as studio-grade and zero-shot, retaining the speaker's delivery style without a separate training step. Its stated focus includes teeth, beard detail, and lip texture, which makes a tight shot useful for evaluation.
Call sync/lipsync-2-pro/lip-sync with a video URL, a replacement recording URL, and an optional timing mode. Processing is asynchronous: retain the task identifier and retrieve completion through polling or a callback. The returned media is an MP4 carrying the uploaded audio's voice. There is no voice-cloning step.
Request fields
| Field | Requirement and default | Input and billing role |
|---|---|---|
video_url | Required; no default | Footage showing a clear face. Its duration does not directly set the charge. |
audio_url | Required; no default | MP3 or WAV, 1–300 seconds. Uploaded audio seconds are rounded up for billing. |
sync_mode | Optional; cut_off by default | Choose cut_off, loop, bounce, or remap. The choice leaves audio-second billing unchanged. |
These three fields apply throughout this lip-sync endpoint. No text prompt, resolution setting, or silence mode is accepted. Prepare the recording before sending the request; the endpoint takes finished speech rather than a written line.
Decide how unequal inputs should align
cut_off: end at the shorter input. Longer audio can lose its ending in the returned clip.loop: repeat the video to cover an audio track that outlasts it.bounce: alternate forward and reverse picture playback to cover the longer recording.remap: stretch or compress the video timeline to the recording's length.
Choose by watching the source movement as well as comparing duration. A repeated gesture or reversed motion may be distracting in a dialogue scene. Review the completed MP4 before placing it into an edit.
Billing measures the submitted recording
The charge counts input-audio seconds, rounded up to whole seconds, with audio in the 1–300-second range. Video duration and the selected timing mode do not change that rule. Consult the Pro model page for the current per-second price.
Under cut_off, a shorter result does not reduce the submitted audio duration used for billing. If the audio extends beyond the video and you intend to discard that ending, edit the recording before submitting. A 4.2-second audio file counts as 5 billable seconds under the rounding rule. Content-review refusals are not charged.
Prepare media for a useful comparison
Choose an authorized speaking shot with a face you can see clearly. For the Pro detail check, examine the mouth, teeth, and beard texture at the intended viewing size. Supply a finished MP3 or WAV recording and listen through the words and pauses before measuring its duration.
Keep the initial trial short enough to inspect closely. Compare the input and output while listening to the replacement recording, then examine the ending for a clipped word, a repeated gesture, a reversal, or a speed change associated with your chosen mode.
Watchmaker example and measured results
Our watchmaker-close-detail example uses a tight shot of a watchmaker in an open-collar navy shirt with salt-and-pepper stubble. The source runs 6.04 seconds at 1280×720; the replacement recording runs 4.2 seconds. The request input is:
{
"video_url": "https://cdn.spicyapi.ai/models/examples/inputs/8bce11f8f3cee83e.mp4",
"audio_url": "https://cdn.spicyapi.ai/models/examples/inputs/deddbb8db32c1ca2.mp3",
"sync_mode": "cut_off"
}
In our tests, that input returned 4.21 seconds of MP4 at 1280×720 and 24 fps. The mouth followed the replacement line, while lip texture, teeth, and stubble remained visible. This is a measured short close-up, without a controlled quality comparison against another model.
A separate timing test paired a 4.0-second, 854×480, 24 fps source with 4.55 seconds of speech. With cut_off, the result ended at 4.0 seconds. That observation illustrates trimming behavior; use the audio-second rule above to plan billing.
Content review and file limits
A content filter runs on this tier. Our Mild input processed; two Moderate attempts were refused during processing. Refused jobs incur no charge. These observations do not establish outcomes for untested levels.
Use faces, video, and voices you are entitled to use, and ensure every person shown is an adult. Impersonation is prohibited. Generated files remain available for 14 days, after which they are deleted; download the accepted MP4 during that window.
Compare nearby lip-sync options
Sync Lipsync 2 uses input-video seconds for billing and offers silence in addition to the timing choices described above. Consider it when those contract differences or content-review behavior matter. Pro is the detail-focused choice described by Sync for close-up work.
Sync Lipsync 3 and VEED Lipsync 2.0 are additional model pages to review. Check their individual input and billing rules before testing; the watchmaker observation does not establish a performance ranking among them.
API quick start
# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $SPICY_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "sync/lipsync-2-pro/lip-sync",
"input": {
"audio_url": "https://example.com/reference.mp3",
"sync_mode": "cut_off",
"video_url": "https://example.com/reference.mp4"
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.


