Skip to content

MNSS (Music Noise Segmentation on a Spectrogram) is a deep-neural network based preprocessing technique that pre-filters unnecessary noise. MNSS is based on the convolutional neural networks and uses softmax value as a probability of noise existence.

Notifications You must be signed in to change notification settings

tango4j/music-noise-segmentation-on-a-spectrogram

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

music-noise-segmentation-on-a-spectrogram

  • Goal : An effective way to classify and analyze audio signal with deep neural networks.

  • The whole project is based on the Keras(Theano)

  • Python dependencies : matplotlib, scipy, numpy, pylab, librosa, pickle, scikits, keras, skimage

  • Must downloads : The learned weight of neural networks; Save this file into "pickel_folder"

https://www.dropbox.com/s/ivp28e2x6gp8hfx/NNM_cla_ss_0__music_noise__j26_0000000000002_SR11kHz_goodmare.pickle?dl=0

  • This technique is investigated for the purpose of making a preprocessing system for music information retrieval (MIR) techniques.

Note : As you can see in the above picture, the proposed convolutional neural networks successfully segments a spectrogram as their identity. For detailed information, refer to my paper that is presented in LBD session in ISMIR 2015. http://ismir2015.uma.es/LBD/LBD27.pdf

Ask any question : inctrljinee at gmail.com

About

MNSS (Music Noise Segmentation on a Spectrogram) is a deep-neural network based preprocessing technique that pre-filters unnecessary noise. MNSS is based on the convolutional neural networks and uses softmax value as a probability of noise existence.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages