GitHub - barrelorgandiscovery/fftmusings: Project applying neural networks to audio data, in particular, attempting to train an LSTM to recognize and reproduce music based on raw music samples. Uses FFT/Phase Vocoder for input followed by an autoencoder and then an LSTM with a mixture density cost function to model the sequence of data.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 10 Commits
data/daa		data/daa
src		src
.gitignore		.gitignore
pom.xml		pom.xml
processInput.sh		processInput.sh
recover.sh		recover.sh

About

Project applying neural networks to audio data, in particular, attempting to train an LSTM to recognize and reproduce music based on raw music samples. Uses FFT/Phase Vocoder for input followed by an autoencoder and then an LSTM with a mixture density cost function to model the sequence of data.