Sarangi

Sarangi is a Music Classification Application that can classify mp3 files based on Genre and Mood. The feature extraction necessary for the classification is also part of this project.

Compilation

Assuming mvn is installed on your machine, just move into the App folder and run mvn compile

cd App/
mvn compile

Creating a jar file

In the App folder, run mvn package assembly:single

mvn package assembly:single

Running the application

Assuming you have all necessary jar files in target/, you can run one of four commands:

• extract: Extract features of songs in given folder.

java -cp ".:target/*" com.sarangi.app.App extract -F SongFolder -f features.txt

• train: Train a classifier from the extracted features.

java -cp ".:target/*" com.sarangi.app.App train -f features.txt -c classifier.txt -C SVM -l 0

• Option "-C"
  • SVM: Support Vector Machine
  • ANN: Artificial Neural Network
• Option "-l"
  • 0: Genre
  • 1: Arousal
  • 2: Valence

• test: Run K-fold validation on the given song features.

java -cp ".:target/*" com.sarangi.app.App test -k features.txt -C SVM 

• classify:

java -cp ".:target/*" com.sarangi.app.App classify -c classifier.txt -C SVM -f awesomeSong.mp3

Or just type --help

java -cp ".:target/*" com.sarangi.app.App --help

Running the jar

If you have created the jar.

java -jar App-2.0-jar-with-dependencies.jar