hu-accompany

Ramus execpted recieve: (We compare to multiple good recording)

WE WILL AVERAGE SAMPLE RECORDING UP FOR AN OVERALL SAMPLE RECORDING

Good_recording: [{notevalue:int(hz), duration:int(s), volumn:int(0~1), ...}]

User_recording: [{notevalue:int(hz), duration:int(s), volumn:int(0~1), ...}]

Rating will be =

  factor1\*abs(good_recording[index][notevalue]-user_recording[index][notevalue]) + factor2\*abs(good_recording[index][duration]-user_recording[index][duration]) + ... + factorsomething\*abs(user_recording[index][duration]-user_recording[index-1][duration]) + factoranotherthing\*abs(user_recording[index][duration]-user_recording[index-2][duration])+ ...

Concerns about rubato:

Check if the tempo is decreasing / increasing in sequence

Next step: learn reinforcement learning

Testing music: Accolay Concerto one in a minor

Get the machine learning out of here we are just random high school student who vibecode the entire project...

Built With

• c
• c++
• cmake
• dart
• html
• objective-c
• python
• ruby
• rust
• shell
• swift