How to Make a Lip Sync AI Music Video

You can create a lip sync AI music video in literally minutes. No editing, no After Effects, no complicated workflow. However if you want it to look good (and not “AI weird”), there’s a specific process that consistently gets me better results.

In this guide, I’ll show you how to make lip sync AI music videos using OpenArt’s Singing Video feature. I’ll be comparing the best image and ai lip sync video models side-by-side and will give you a lot of practical advice along the way (tips and tricks I learned after creating dozens of Music Videos using OpenArt).

Check out my YouTube video for a full breakdown and side-by-side comparison of AI lip sync models (OpenArt Lip Sync vs Hedra vs Kling vs Creatify Aurora)

Navigation

What You’ll Learn

The 30-second workflow (the “it actually works and doesn’t look bad at all” proof)
The proper workflow (what you should do if you care about quality)
The best image model to use, depending on the style of your video (realistic, animated, …)
The best video model to use for creating AI lip sync music videos (Kling vs Hedra vs OpenArt Lip Sync vs Creatify Aurora)
How to avoid common mistakes when reviewing your Storyboard within OpenArt
How to preview, fix, and export your final AI lip sync music video cleanly

Create AI Lip Sync Music videos: 30 second workflow

If you just want to test the tool fast, here’s the speed-run:

Go to OpenArt
Click on Create Music video
Choose Singing Video
Upload your song
Pick (or upload) a character
Open Advanced Options
Hide anything you don’t need
Select an image and video model you like
Click Create

But here’s the truth: the “speed method” is for quick wins. If you want a video you’d actually upload confidently, here’s the proper workflow:

The Best Way to Make a Lip Sync AI Music Video (Step-by-Step)

This is the workflow I recommend if you care about clean mouth movement, stable faces, and fewer “uncanny” moments.

Step 1: Create a Singing Video in OpenArt

Go to OpenArt → Create Music video → choose Singing Video

Why Singing Video?

It’s built specifically for lip sync AI music videos
It’s more “music-video-first” than generic video generators
They’re the best on the market for one-click music video solutions.

Step 2: Add Your Song

You can:

Drag & drop an audio file
Or paste a link (directly from SUNO for example, if that is the AI music generator you are using)

You can also trim the song if needed, and OpenArt supports songs up to 5 minutes

Quick tip: If your goal is testing lip sync quality, use a short clip first (10–30 seconds). Faster iteration = better results.

Step 3: Choose the Right Character for Your Vocals

Choose a character that matches your vocals. This sounds obvious, but it affects how believable the openart singing video feels.

Choose character for Lip Sync music video

You can pick:

Your own custom character (best for uniqueness)
- You can make your own characters by uploading a Photo OR by making a character using Openart’s “Character” feature (with this method your character will show up under the “My Character” section and useable for any future video you make)
A community character (fine for quick experiments)

Pro move: If you’re serious about consistency, create your character with OpenArt’s character feature first, then reuse it for multiple videos.

Outfit Changes: Yes or No?

If you enable outfit changes, your video can feel more dynamic, but it also introduces variability. I’ve personally never really ran into issues when having this option selected. You can always edit outfits once you get to the storyboard.

Step 4: Set the scene (Optional)

You can describe a scene… but you don’t always need to.

OpenArt uses your:

Song (it interprets the style and mood of your song)
Lyrics (it pays attention to your lyrics and what the song is about)
Character

…to decide the visuals.

If you do type something here, a good start is to keep it simple and visual (not poetic). Examples:

“Studio performance, soft spotlight, close-up singing shots”
“Cinematic stage, shallow depth of field, subtle camera movement”

This is all about experimenting.

Step 5: What Visual Style to choose?

If the character you have selected is photorealistic and you want a music video that is cinematic and realistic:

✅ Match Character

This keeps the whole video aligned with your character’s look, instead of forcing a random style across everything.

Of course you can experiment with other styles (and you should :D), but this once again introduces variability.

Aspect Ratio

16:9 for YouTube
9:16 for Shorts/Reels

Advanced Options (This Is Where Quality Happens)

Advanced Options for Lip Sync AI Music videos

Show B-roll shots in music video?

If you want a strong lip sync AI music video, you usually want the camera on the singer. However, this really depends on how long the song is you have uploaded. If it is short, I personally prefer to turn this option off and keep the cinematic shots to a minimum. If you’re using a 3 minute + song, then by all means, B-roll often can work really wel!

So for short songs (or lip sync tests):

✅ Turn B-roll OFF

B-roll adds cinematic shots without your character, which weakens the lip sync focus.

Keep “Match to Beat” ON

If the option is available:

✅ Leave match to beat enabled

It helps the visuals feel rhythm-aware, which makes the final video feel more like a real music video.

Captions

If you don’t need captions:

❌ Turn them off

If you do use captions:

Always review them (AI captions can be wrong)
Edit mistakes inside the timeline editor once you’ve generated your full video
You can also delete all the captions later on. So there really isn’t any “bad” option to choose here

Best Image Model for Lip Sync AI Music Videos

Some image models can mess up photoreal faces, especially eyes.

If you’re going for photoreal:

Use models that produce stable eyes + clean skin detail

Rule of thumb: If the still image has uncanny eyes, the video will usually amplify it.

Best AI Lip Sync Video Model

You’ll typically see multiple video models depending on resolution.

At 1080p, you may have fewer model options
- At the time of writing this blog, only KLING Avatar Pro is available here. This is a good option but it will HIGHLY depend on your song choice. I notice that it often makes the character smile, which might not be a great thing in some cases. (Check out my side-by-side comparison at the 13 minute mark of the YouTube video I’ve posted above)
At 720p, you usually get more choices
- OpenArt Lip Sync, Hedra and Kling Avatar cost the same credits. Only Creatify Aurora is significantly more expensive. Both Hedra and OpenArt Lip Sync are very solid options if you want Photorealistic result. I wouldn’t recommend Kling, unless perhaps if you’re chasing a more creative or animated music video. (Again, I advice you to check out my Youtube video for a comparison of these models)

Pro move: Don’t chase 2K early. Start with default quality, nail the look, then upgrade resolution if needed.

The #1 Tip: ALWAYS Preview the Storyboard

✅ Preview Storyboard first

It doesn’t cost extra
You’ll spend the credits anyway when generating the full video
But storyboard preview lets you catch issues early –> It is a lot cheaper (in terms of credits) to change an image rather than having to change the image AND the video

How to Review the Storyboard on OpenArt

When the storyboard is generated, you’ll see:

An image description
The generated image
The prompt OpenArt will use to animate this image into a video

Photoreal Checklist

When making realistic AI music videos. Always check the following:

Eyes (sharp, aligned, not wonky)
Teeth/mouth area (no weird shapes)
Hands (if visible)
Anything “too magical” (breath smoke, strange glow, random objects)
Nothing “weird” going on in the background of the image

Fixing OpenArt’s “Poetic Prompts”

For some reason OpenArt often writes these prompts like a novel… Beautiful words, but not helpful for video motion. (Granted, the Story Mode option is still in Beta so this could easily be fixed in a future update)

For better results:

Remove overly poetic and descriptive lines
Keep only what matters… I usually like to keep this focused on just camera motion

Example approach:

Delete: “eyes shimmering toward the lamplit listener…”
Keep: “slow push-in, subtle handheld, close-up, singer facing camera”

KEY ADVICE: The simpler you keep this prompt, the less likely you will have to regenerate the video because of some funky things that might be happening 🙂

Fix One Bad Clip Without Regenerating Everything

Once you’ve reviewed the storyboard and generated the full video, you can now change clips individually. If one scene looks weird, you can usually:

Identify the bad clip
Retry only that clip
Or fix the image first, then retry the video

This saves credits and keeps your workflow fast.

Editing After Generating your AI Music video: Timeline + Captions

A feature many people miss: you can review and adjust the timeline (you will find this option at the top of your screen).

You can:

Drag clips to make them longer/shorter
Add captions (and edit them when the AI makes mistakes)
Add your own media if you want to

Tips for Better AI Lip Sync Music Videos

If the lip sync looks off:

Use a cleaner vocal track (less heavy effects)
- Many AI Music generators also offer the option to download just the vocal tracks. This can be a good option too for better AI lip sync accuracy. The trick is to then download the video and sync it with your actual music using a video editor like Capcut for example. Check out my begginer-proof video on how to make music videos with Capcut.
Avoid ultra-fast syllable delivery for your first tests
Use close-up images and camera prompts (mouth detail significantly improves)

If the face looks uncanny:

Switch image models
Retry the image where eyes look wrong
Remove “poetic prompt clutter” and keep camera motion only

If the video has one cursed clip:

Retry only that clip
Replace the image, then retry again

FAQ

How do I make lip sync AI music video with OpenArt?

Use OpenArt → Create → Singing Video, upload your song, choose a character, select an image model + lip sync model, preview the storyboard, fix bad frames/prompts, then render and export.

What’s the best lip sync AI tool inside OpenArt?

OpenArt’s own OpenArt Lip Sync model is a reliable default for most songs. The “best” can vary by song style, so test 10–15 seconds with two models and pick the winner before your full render.

Can I make an OpenArt singing video with my own character?

Yes. You can upload your own character (I would recommend an image generator like Google Nano Banana for this, which is also available within OpenArt), or create one using OpenArt’s character feature (also using the Google Nano Banana option for better results), then reuse it for future singing videos.

What other options are there for creating music videos?

I have created mutliple blog posts and videos covering different methods on how to make music videos:

The fastest way, using beautiful visuals that sync to your music: How to make a visualizer for music videos
The creative way for making professional looking music videos: CapCut Lyrics Music Video Tutorial

Wrap-Up

That’s the full workflow for how to make lip sync AI music video the right way: fast enough to be fun, but controlled enough to get results you’d actually publish.

If you try this, drop a comment on my YouTube video with what worked best for you, especially which lip sync AI tool and model combo gave you the cleanest result.