MANNER icon indicating copy to clipboard operation
MANNER copied to clipboard

Published 20 hours ago • verified publisher icon winddori2002

customized evaluation

Open catalwaysright opened this issue 1 year ago • 2 comments

Hi! Is it possible to input our own audio and get the enhanced audio through the model?

catalwaysright avatar Aug 21 '22 23:08 catalwaysright

Hi!

We have uploaded the code for custom enhancement. You can find details in './custom_enhance.py'. The code is only for inference, and if you want to evaluate, you can refer to './eval_measure.py'.

Thanks for your interest.

winddori2002 avatar Aug 29 '22 08:08 winddori2002

Hi! I have adapted manner into a custom audio processing demo. demo.py.txt attached is the source code.

Because I dont really understand how manner works I have fed it only 1 or 2 seconds at a time of audio also because my script wont allow for a greater delay

manner, as other networks and methods of noise gating, only has a short term memory(seemingly) for variations in amplitude(as in a noise gate approach) and cannot(as used here) significantly attenuate noise outside of windows of activity

falseywinchnet avatar Sep 10 '22 04:09 falseywinchnet