I built an open-source music video generator

YinTokey — Thu, 07 Aug 2025 15:23:30 +0000

1. Introduction

I enjoy listening to background music on YouTube and have tried creating similar music using tools like Suno AI, LeonardoAI, Runware, DALL·E, and CapCut. I noticed that these music videos usually follow a few basic patterns:

Background image
Music (instrumental or with vocals)
Visual effects (optional)
Animation (optional)

With my experience using FFmpeg for video generation, I decided to build a tool that can create these music videos from a simple prompt, without any video editing work.

Here’s how the tool works:

2. Background Image Generation

We start by entering a few keywords. The tool uses ChatGPT to enhance your prompt, and you can upload a reference character image if you want. For example, I used an AI influencer image I found on Google.

This tool enables both instrumental music and music with a singer (so you can create AI influencer music videos).

Next, the tool calls Runware AI to generate background images. Runware is an image generation service, similar to Midjourney.

3. Music Generation

The next step is music generation. ChatGPT analyzes the background image and crafts a better music prompt. If the character is female, the generated music will feature a female singer.

Then, the Suno API is used to generate the music file.

4. Video Generation

Once we have both the background image and the music file, the tool creates a video by using the image as each frame, just like a video editing tool would.

Finally, we use FFmpeg to generate the finished video.

That’s the entire process.

5. Cost

There are two types of videos. If you use a reference character, the cost is about 12 cents per video. If you use the Juggernaut model for image generation, the cost drops to around 8 cents.

6. Animation (Future Work)

So far, the tool generates videos with static images. But what about animated images? I haven’t implemented this feature yet, but Leonardo.AI’s Motion One model can generate animated videos by adjusting the movement strength (I set it to 1 in my test).

Below is an example of an animated video generated using Leonardo.AI:

Leonardo.AI also supports API calls, so integrating animation into the project is definitely possible in the future.

7. Project Link

Here is the project link, hope you like it!

https://github.com/YinTokey/AutoMV

DEV Community: YinTokey