Freesound

• This project is to solve a kaggle challenge ( https://www.kaggle.com/c/freesound-audio-tagging/ ) : Build a model which can learn 9000 audio files classified into 41. The model should predict when a new file from this 41 is supplied
• Training set was in wav format sound files, suitable python packages are used to convert these into keras requisted format
• Appropriate metrics and suitable benchmark methods to measure each learning iteration and tuning the model to improve accuracy.
• An application which can understand and classify sound file could be implemented for various use cases.