send audio

https://tenor.com/view/fade-emoji-funny-gif-25240293

this model is mid in my opinion, I should not have mixed the two Foxys

is the reverb included in the dataset?

I tried to exclude as much reverb from the original foxy's voice as I could but it's not perfect

ah

oh yeah its barely dereverbed

yeah it didn't work well :[

what software did you use to dereverb?

https://mvsep.com/en/home

it has a dereverb thing with different options

hmm which dereverb separation model?

i know uvr has its own dereverb

my settings I usually use

oh

try this instead

echos are like reverbs but with longer delays

hmm odd

you can try bandit plus since that mostly takes out most of the effects

what's that?

another thing could be because you set the output encoding to mp3 and not wav

not sure what mp3 output does but hearing the dereverb version didn't change that much

its built for normal speech

i think sending the datasets might be breaking the server's rules so is it okay if we add each other on Discord?

I changed it to wav on deecho dereverb and it sounds different

nice!