You generated a video clip in Runway or Pika, generated voice-over separately in ElevenLabs or OpenAI TTS, dropped them onto the same timeline, and the mouth movements miss the audio by anywhere from a fraction of a second to obvious vowel mismatch. The eye catches lip sync errors at thresholds as tight as 80 milliseconds. The fix depends on whether you can re-render with sync, sync afterwards with a dedicated tool, or compose around the mismatch.
Common causes
Ordered by what we see most often.
1. Audio and video generated independently
The most common case. You wrote a prompt like “woman speaking to the camera,” generated 5s of video, then generated 5s of audio. Neither model knew about the other. The mouth movements in the video are pure speech-like motion, not aligned to specific phonemes.
How to spot it: Did you generate the video clip without uploading or referencing the exact audio file? If yes, this is your case.
2. Tool does not support real lip sync
Runway Gen-3 does not align mouth to audio. Pika 1.5 has a Lip Sync feature but only for clips it generated, not for arbitrary footage. Kling has a Lip Sync mode introduced in 2.0. Sora has no lip sync. If you used a tool that does not support it, no amount of tweaking the prompt will fix it.
How to spot it: Check the tool’s docs for “lip sync” or “audio-driven motion.” If absent, the tool is the bottleneck.
3. Frame rate mismatch between video and audio
You generated video at 24fps and the audio waveform is being mapped against a 30fps timeline (or vice versa). The mismatch compounds over time — the first second looks fine, by the fourth it is half a phoneme off.
How to spot it: Open the video in your editor and check the fps. Compare with the audio file’s sample rate (44.1kHz / 48kHz is normal). Then check the project sequence settings.
4. Audio has silence padding at the start
ElevenLabs and OpenAI TTS sometimes prepend 50-200ms of silence to the output. If you snapped the audio to clip start, the spoken portion is now offset and lip sync is misaligned by that padding amount.
How to spot it: Zoom into the waveform at the start. If there is flat audio before the first phoneme, that is your offset.
5. Stretched / time-remapped audio
You sped up or slowed down the audio to fit the video duration. Lip movements in the video are at the original speed; remapped audio is not. Sync drifts proportionally.
6. Talking-head shot generated with wrong “speaking” cues
Some video models produce mouth-open-mouth-closed motion that maps onto vague speech but never matches specific words. The mouth shapes for “M”, “P”, “F” (lip closures) are missing entirely.
Before you change anything
- Save both source assets (video and audio) at their original quality.
- Note the exact tool, model, and version used to generate each.
- Decide how important sync is: a brand explainer needs tight sync; B-roll voiceover does not.
- Confirm the frame rates and sample rates of both assets match your edit timeline.
- Commit or back up the current edit before re-rendering — re-generation burns credits.
Information to collect
- Both source files, original quality.
- Frame rate of the video, sample rate and codec of the audio, project sequence settings.
- Whether the audio includes leading silence padding.
- Transcript of the audio with timestamps (most TTS tools export this).
- A specific timestamp where the mismatch is most obvious.
Shortest path to fix
Step 1: Decide your sync strategy
Three legitimate paths:
- End-to-end lip sync in one tool: best quality, smallest control over voice / look.
- Generate video and audio separately, then sync afterward with a dedicated tool. Best control, most steps.
- Compose around the mismatch: cut away from the face during words that obviously break sync.
Choose based on the use case.
Step 2: For end-to-end lip sync, use a tool that supports it
Current options (2025-2026):
- HeyGen: avatar-based, generate avatar speaking your script directly. Best out-of-the-box.
- D-ID: similar — upload a portrait, give it a script, get a talking-head video.
- Synthesia: avatar library + voice cloning + script-to-video.
- Pika Lip Sync: only on clips Pika generated; upload audio + select clip.
- Kling 2.0 Lip Sync: works on a wider set of inputs.
- Runway Act-One: 2025 feature, drive a generated character with a reference performance video.
For brand work where lip sync must be tight, generate the entire shot in one of these tools rather than stitching separately.
Step 3: For separate generation, sync with a dedicated tool
After you have video and audio separately:
- Import both into a tool like Sync.so (formerly Wav2Lip cloud), Synthesia Lip Sync, or a local Wav2Lip / SadTalker pipeline.
- Upload the video clip + audio file.
- The tool re-renders the mouth area aligned to your audio.
Wav2Lip is open source; Sync.so charges per minute but produces stronger results on natural-looking faces.
Step 4: Match frame rate and trim silence
Even with the right tool:
- Convert all clips to 24fps (cinema standard) or 30fps (web standard) — pick one and stick to it across the project.
- Trim leading silence from audio before placing on timeline. Use a tool with a “silence detection” feature (Premiere, DaVinci, Audition all have it).
- Lock audio sample rate to 48kHz (video standard) — re-export TTS at 48kHz when possible.
Step 5: For the compose-around path
If you cannot regenerate or sync:
- Cut away from the face during the worst mismatched moments. B-roll, product shots, environment cuts — anything that hides the lip mismatch for 1-2 seconds.
- Use over-the-shoulder shots where the mouth is angled away.
- Use lower-third graphics during quoted text.
- For social-format edits (TikTok, Reels), captioning the speech draws viewer attention from lip mismatch.
Step 6: Re-record the voice with cadence matching the video
If you have voice control (ElevenLabs Studio, OpenAI TTS), re-render the voice trying to match the cadence of the video’s mouth movements. Insert short pauses or speed up sections to align speech with mouth opens.
How to confirm the fix
- Play at full speed with sound. The eye should not catch the mismatch.
- Play at 25% speed. Individual phoneme alignment should be within ~80ms.
- Show the clip to someone unfamiliar with the project. Ask whether anything feels off.
- Check a clip from the middle and the end of the video — sync drift compounds, so the end is the hardest test.
If it still fails
- Reduce to the smallest reproduction: just the 2-second segment where the mismatch is worst. Most “the whole clip is off” complaints collapse to one segment.
- Try the alternate path — if you tried end-to-end and it was rigid, try separate-then-sync. If you tried separate-then-sync, try a different post-sync tool.
- For business-critical content, consider an avatar tool (HeyGen, Synthesia) even if you wanted a more cinematic look — sync wins.
- Package the source video, source audio, edit timeline, and the bad moment before asking community help.
Prevention
- Decide before generating whether lip sync matters; pick tools accordingly.
- Standardize project frame rate (24fps cinema or 30fps web) across all generations.
- Use 48kHz audio throughout to match video standards.
- Build a “sync workflow” doc per use case (brand video → HeyGen, B-roll → separate-then-sync).
- Add 200ms of buffer at each cut point so minor sync drift hides behind cuts.