COSPA

Complex-valued Spatial Autoencoders for Multichannel Speech Enhancement

In this repository you will find exemplary results from our yet-to-be-published paper Complex-valued Spatial Autoencoders for Multichannel Speech Enhancement.

In addition, you can find the implementation code, using Pytorch, of the proposed COSPA.

Audio examples

Instructions

You can download individual audio examples by clicking on the respective download buttons or by using left click then save link as.

Example 1

Unprocessed microphone signal	Single-channel CDNN	DNN-MVDR	DNN-MWF	OMVDR	COSPA	OGMVDR

Example 2

Unprocessed microphone signal	Single-channel CDNN	DNN-MVDR	DNN-MWF	OMVDR	COSPA	OGMVDR

Example 3

Unprocessed microphone signal	Single-channel CDNN	DNN-MVDR	DNN-MWF	OMVDR	COSPA	OGMVDR

Example 4

Unprocessed microphone signal	Single-channel CDNN	DNN-MVDR	DNN-MWF	OMVDR	COSPA	OGMVDR

Example 5

Unprocessed microphone signal	Single-channel CDNN	DNN-MVDR	DNN-MWF	OMVDR	COSPA	OGMVDR

Example 6

Unprocessed microphone signal	Single-channel CDNN	DNN-MVDR	DNN-MWF	OMVDR	COSPA	OGMVDR

mdiligent/COSPA