Inspiration
My mother always listens to music on YouTube, but they would always in a song compilation format, leaving her having to search for each individual song and adding it to her playlist. I wanted to make sure she wouldn't have to manually search, nor have to be connected to the internet.
What it does
It takes the thumbnail of any YouTube video, uses OCR to output a file of the words in both English and Chinese, then parses it into a specific format using Gemini. Gemini then filters it and matches a YouTube video close to the song and uses an unofficial YouTube API to download the video locally as an MP3.
How we built it
We used Python for the backend CLI. We used yt-dlp, Gemini API, and pytesseract.
Challenges we ran into
We had difficulties directing PyCharm to the correct translator that allowed for the inclusion of libraries such as Tesseract. We also had challenges with not being able to run pip install directly in the python terminal.
Accomplishments that we're proud of
We are proud to have made a project that others could benefit from.
What we learned
We learned how to use API keys and to code in Python to some degree.
What's next for Song Ripper
We are planning on adding more features to it, such as certain text highlighting, custom picture insertion, and a more refined UI.
Log in or sign up for Devpost to join the conversation.