Goal:
Results:
Goal:
- Confirm plans for final demo with Richard and team
- Clean up github and further documentation
Results:
- Final demo plans confirmed
- Use LSTM neural network and Nottingham dataset to train model to generate melodies
- Use said melodies to combine with a given chord progression
- Correct the combined melody and chord progression using Adam's note correction tool
- Explore different features of neural network architecture to determine optimal features
Next steps:
- Execute final demo plans and prepare presentation
Notes:
- Did not meet; Thanksgiving Break
Notes:
- Did not meet; Richard out of town for presentations
Goal:
- Prepare 3/4th presentation
- finish out How to Generate Music using a LSTM Neural Network in Keras and incorporate our code into the LSTM neural net
Results:
Next steps:
- Begin planning for final demo
Goal:
- make more progress on How to Generate Music using a LSTM Neural Network in Keras
Results:
- Began looking into the codebase and thinking of ways to incorporate our current code into a LSTM neural net
- Thought of new ways to make our current code more robust
Next steps:
- Finish out reading through the article and code and incorporate ours
- prepare for 3/4ths presentation and begin thinking about final demo
Goal:
- Clean up last week's code, add documentation and README
- Read through article about How to Generate Music using a LSTM Neural Network in Keras
Results:
- Code ready to present
- Roughly halfway done with LSTM article
Next steps:
- continue reading How to Generate Music using a LSTM Neural Network in Keras
Goal:
- combine chord text parser with Adam's rules about note correction to produce corrected note stream
Results:
Next steps:
- clean up documentation for github
- begin deep learning implementation with resources Richard sends
Goal:
- Write python method that takes in strings of text representing a chord’s name “C:maj”, and returns the notes that comprise it
Results:
Next steps:
- combine chord text parser with Adam's rules about note correction to produce corrected note stream
Goal:
- Write python method that takes in strings of text representing notes and returns chord progression
Results:
Next steps:
- write method to essentially do the reverse: given a string of a chord’s name “C:maj”, return the notes that comprise it
Goal:
- Learn more about RNN/LSTM and look for actual code
Results:
Next steps:
- Continue learning about CNN’s and deep learning topics
Goal:
- Learn from other group members about CNN's
- Teach group members about LSTM & music21
- Discuss potential UI designs
Results:
- Everybody generally caught up to speed (at each other’s level of understanding) on deep learning topics
Next steps:
- Continue learning about CNN’s and deep learning topics
Goal:
- Figure out which subproject I will be focusing on
Results:
- Will be working on reMelodize and reHarmonize; focusing on chord progression RNN
Next steps:
- Read into topics regarding LSTM and look into music21