Rap Battler

We used image and speech generation and cloning to make Micheal Buffer as our chosen rap battle announcer.
This is our simplistic front end which receives two contestant names.
We used Respell to automate the image and script generation process.
We used D-ID to animate the images of the chosen celebrities.

Inspiration

We were inspired by the growing interest in AI-powered content generation. We wanted to create something that was fun and innovative, while showcasing the abilities of different emerging technologies. , Rap Battler allows you to witness the rap battle you always wanted to see. With ChatGPT, Mid Journey, and Eleven Labs, we wanted to create a truly unique and entertaining experience.

What it does

Rap Battler is a generative AI application that generates a rap battle between your favorite celebrities. With Rap Battler, you can select your preferred celebrities and witness a rap battle showdown. The platform generates a fully-formed rap battle script that can be shared on social media or used for entertainment purposes. Rap Battler is a fun and innovative way to explore the capabilities of AI-powered content generation.

How we built it

Rap Battler was built using ChatGPT, Mid-Journey, ElevenLabs, and React. We started by using ChatGPT to generate the rap lyrics, which allowed us to generate humorous content very quickly. To make the celebrities look and sound as though they are rapping, we used Eleven Labs to generate the celebrities' voices and Mid Journey's API to generate images of the celebrities, and D-ID to make these images move as though the celebrities are speaking. Finally, we used React to build a simple front-end that takes in two contestant inputs and outputs the generated rap battle. Overall, Rap Battler was a fun and collaborative effort that allowed us to explore and implement these amazing technologies in cool and entertaining ways.

Challenges we ran into

At first, the idea seemed simple to implement, since the content generation was already provided in existing technologies. However, we quickly discovered that integrating these different technologies to produce a coherent rap battle posed a lot of challenges. We had a lot of difficulties getting these technologies to produce exactly what we need. We found that images and videos generated by AI doesn't always look appealing, sometimes not even recognizable. Feeding the script from ChatGPT to Eleven Labs to generate rap audio posed a significant challenge, because we had to parse the script to clarify who was speaking. Overall, the integration of these technologies to get them to work together was the biggest challenge of this project.

Accomplishments that we're proud of

We are very proud of the project we created, the ideas we came up with, and our fun and collaborative effort. This was a very fun project and we learned a lot from working with all these different technologies.

What we learned

We learned how difficult it is to integrate different AI technologies together. Though content generation AI is very powerful, it isn't always going to produce what you want or need very easily. Finally, making different AI's work together coherently in a project can pose many unexpected difficulties.

What's next for Rap Battler

Our team is looking forward to exploring further with the different tools that we used in this Hackathon. Going forward, we hope to fully implement automated functionality.