search

AkojimaSLP / Neural-mask-estimation

37 stars 9 forks source link

Beamformed Samples Missing #3

Open dreamibor opened 4 years ago

dreamibor commented 4 years ago

Hi @AkojimaSLP , When I am exploring the repo, I found that the number of input samples and output samples are different, for example, if I got a 50-second wav audio (800000 samples) and the output will be only 799744 samples. I see it's related to the MVDR beamforming part, do you know why it's happening? Thank you very much!

dreamibor commented 4 years ago

Update: a straight-forward example is the beamformed audio under the /result folder, enhacement_all_channels has file size of 205,356 bytes while the original wav file such as F02_011C021A_BUS.CH1.wav has 206,114 bytes)