Inspiration
For an assignment in our AP Environmental Science class, we had to make a rap song/cover and perform it in front of the class. Despite finding perfect songs that matched our message and style, we couldn't locate instrumental versions to perform with. We wanted to create a web app that would address this issue, as well as being able to easily display the karaoke so people could have fun by themselves or with a group of friends.
What it does
KaraokeKit transforms any song into a karaoke track in three simple steps. Users can either upload audio files or paste YouTube links. Our platform then separates the vocals from the instrumentals and transcribes the lyrics. Once processing is complete, users can sing along with just the instrumental track while following synchronized lyrics on screen. The intuitive interface works seamlessly across devices, featuring light and dark modes for comfortable viewing.
How we built it
- Frontend: SvelteKit for a responsive and reactive UI, with Tailwind CSS for styling
- Backend: Flask API running Python for audio processing
- Audio Separation: The audio-separator library to split vocals from instrumentals
- Lyrics Generation: OpenAI's Whisper model for accurate transcription with timestamps
- YouTube Integration: yt-dlp for extracting audio from YouTube videos
- Deployment: Configured for easy deployment to Vercel (frontend) and any Python-compatible server (backend)
What's next for KaraokeKit
We want to add a recording and a playback feature in the future, and also a 'saved playlists' feature to keep all of your karaoke songs in one place!
Log in or sign up for Devpost to join the conversation.