Bruno

Background music is a 3 Billion Dollar Industry. Bruno just killed it.

Inspiration

A lot of creators get hit with copyright issues when adding music to commercials, YouTube videos, and short form content. At the same time, AI generated music is getting really good and will likely outpace traditional workflows for background tracks. I wanted to make it dead simple for anyone to add high quality, copyright safe music to their videos.

What it does

You upload a video with no audio. I send it through Google Cloud Video Intelligence to understand what is happening in the video. That output is turned into a structured music prompt using OpenAI. The prompt is optimized for Suno, which generates the background track. I then stitch the generated audio onto the original video and return the final version.

How we built it

I used Google Cloud Video Intelligence to analyze video content. OpenAI generates structured prompts for Suno. Suno handles music generation. The frontend is built in Next.js and deployed on Vercel. The backend is deployed on Railway. Everything is wired together through API calls.

Challenges we ran into

One big issue was sending large video files between the frontend and backend. I solved this by uploading directly to cloud storage and passing references instead of raw files back and forth.

Prompting Suno well was also harder than expected. I did a lot of testing to figure out how to structure prompts so the music actually matched the tone and pacing of the video.

Understanding what is happening in a video is non trivial. Google Cloud Video Intelligence helped extract labels and context, but I had to experiment to turn that into something musically useful.

Accomplishments that we're proud of

After building it, I showed it to a bunch of people and most of them could not tell the music was AI generated. They actually liked the tracks.

I am also proud that I fully deployed it. It is not just a demo running locally. It is a working product that came out of a hackathon.

What we learned

I learned how to deploy full stack apps using Vercel and Railway. I learned how to properly structure prompts for Suno. I learned how powerful Google Cloud Video Intelligence is for extracting meaning from video.

What's next for Bruno

I want to add more customization so users can tweak the song after it is generated. I want to support more inputs, like short reference music snippets. Long term, I would explore training or fine tuning models specifically for video to music generation.