Audiomatic

Home
Dashboard
Upload
History

Inspiration

Each member of our team comes from uniquely musical backgrounds, ranging all the way from classical piano to electro EDM music. In other words, when walking down Speedway late at night, or getting a burger with our friends, we often imagine - what would be the best background music for this situation?

What it does

Given an audio-less image or video, our application will create thematic background music reflective of the scenery. In addition, there are QOL features such as a dashboard, activity history, and user-friendly homepage.

How we built it

For the database, we used PostgreSQL and Prisma ORM to store the prompts generated from the image/video inputs, and the prompts used to generate the soundtracks. Our backend was created with Flask in Python for endpoints to accessing images/audio/prompts and to integrate with the AI services we used. We processed images and translated text prompts with OpenAI GPT 4o-mini and did the final translation to an audio file with Meta MusicGen. In addition, we used React for the frontend and Tailwind CSS to design and format our webpages.

Challenges we ran into

Setting up Python virtual environments over multiple different computers was difficult, and we also had difficulty connecting our application to our SQL database. We also had our fair share of troubles with CSS and getting the formatting to look the way we wanted it. Fortunately, we were able to work through these challenges by asking for help, both from other people and online forums.

Accomplishments that we're proud of

Given that this was the first ever hackathon for some of the people in our group, we were incredibly proud to have a working product that accomplishes what we aimed to do. Furthermore, we were even able to add extra features, such as a dashboard and previous audio recordings while streamlining the user experience with an elaborate frontend environment.

What we learned

We learned that although each of us may all have great ideas, the ultimate factor that allows a team to produce a great product is teamwork and effective communication. Furthermore, we learned how to use many of the different resources provided by the sponsors of HackTX, such Intersystem's VectorDB and Intel's AIPC.

What's next for Audiomatic

We would like to use the audio produced by Audiomatic to suggest recommended songs from the internet based on a user's input picture/video. Furthermore, Audiomatic would also be greatly improved by gradually learning a user's taste in music and molding audio generation to this taste.