ale/aiexperiments-ai-duet
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update README_CS310.md

Browse Source 
Add instructions to create a MusicModel and how to use a MusicModel to classify sentiments
This commit is contained in:
Avi Vajpeyi 2017-05-06 17:45:26 -04:00 committed by GitHub
parent fb61e542a9
commit 1945e7669c
1 changed files with 16 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
README_CS310.md
View File

@ -12,3 +12,19 @@ The files modified were:
-static/src/style/common.scss -static/src/style/common.scss
-static/src/style/glow.css -static/src/style/glow.css
To create a music sentiment detection model:
Store happy/sad songs in seperate folders.
-Run XXX path/to/folder/with/midi/files
(This will generate a text file with the notes normailized to the C and c minor scales, and represents the notes as numbers.)
-Run createMusicalFeatureSets.py
(This takes the text files and generates musical feature sets, which is the input data of frequencies of notes, and the labels for each set of note frequencies. This is then saved as a pickle, with the testing and training data sepearated)
-Run trainMusicNN.py
(This takes the pickle created and passes the data through a NN. It saves the NN's weights as a model)
To use the sentiment detection model to classify a song according it its sentiment:
-Run usingMusicNN.py
(This takes the sentiment detection model and a midi file. It converts the MIDI file to a feature set of fequencies, which it passes through the trained NN. It prints the classification of the set.)