search

andabi / music-source-separation

Deep neural networks for separating singing voice from music written in TensorFlow
796 stars 149 forks source link

Try this architecture in the speech enhancement for speech recognition? #2

Open sigpro opened 7 years ago

sigpro commented 7 years ago

Can this architecture recover speech from the noisy speech?

andabi commented 7 years ago

@sigpro The same architecture can be used for removing noise from speech or extracting single speech from dialog as well ;) Take a look at this reference paper: https://posenhuang.github.io/papers/Joint_Optimization_of_Masks_and_Deep%20Recurrent_Neural_Networks_for_Monaural_Source_Separation_TASLP2015.pdf

sigpro commented 7 years ago

@andabi Thank you for reply,will you try multi-channel for better performance?

andabi commented 7 years ago

@sigpro yes i'll it if have some dataset ;) Do you know any dataset that has multi-channel ?

sigpro commented 7 years ago

@andabi I think CHiME challenge's dataset can be used.

ghost commented 7 years ago

Just to extend this question. How do I modify the code to make this work with any dataset? Or is the code specifically created for singing voice separation? Would be awesome if this is a universal solution to many source separation tasks.