Welcome! Let's go straight to the challenge.

📂 You are given the following DATA:

• 🎺 A dataset of musical instruments.
  • The sounds are partially labeled: some instruments have a class label, some don't.
  • The labels are noisy: they are inconsistent and potentially wrong.
• 🎼 A song without instruments.
  • The song has the notes written in, but its instruments are muted.
  • The instrument labels appearing in the song are also given. But we don't tell you which note corresponds to which instrument.

✅ Here's your list of TASKS:

• Label consistently all the instruments in the dataset.
• 🏁 Be sure to look up, as we will have an automatically-update live leaderboard!
• Play the "unknown.mod" song.
• The song is famous! What's the title?
• Generate new sounds and use them to play the song.

You can address the tasks in any order you prefer.

📄 Submit a brief one-page REPORT (don't overdo it!) of your best submission before June 2nd 2024 at 23:59 via email to Prof. Rodolà.

• 🏷️ You'll get points depending on your labeling accuracy.
• 🎶 You'll get +0.25 points if you play the song at least once.
• 🤩 You'll get +1 point if you guess correctly the song title.
• 🎧 You'll get points depending on how we like your song with new sounds.

Here you can find a primer notebook on audio processing.

Make sure to understand everything, as the functions described in it may come in handy during the hackathon.

Here you can find all information regarding the data that you will need to use in the hackathon.

You can now start working on the hackathon!

🏁 Be sure to look up, as we will have an live leaderboard!