Uberduck

I'm just trying to make it exist so I can clone my voices

another version meaning you'll use the mlp voices then? if so what will you use now?

I don't need to waste more time redoing the encoder step

I can just convert the label.json files to metadata.csv

else it'll waste neurons learning when to waste how much time

I thought the audio files were already fine

the one notebook that i've seen use this has a silence trimmer

just copy the one i have, recommended settings and all

Okay, and run the portions that relate to the MLP dataset?

-or no sorry i meant copy just the silence trimmer

It's a package, I think. Don't see anything code-wise.

but there's a bit of code that uses the package

```py
print("Trimming silence...")
!mkdir /content/tempwavs2
for i in tqdm(list(os.scandir("/content/tempwavs"))):
  !sox {i.path} {os.path.join("/content/tempwavs2", i.name)} silence 1 0.05 0.1% reverse silence 1 0.05 0.1% reverse
!rm -rf /content/tempwavs
!mv /content/tempwavs2 /content/tempwavs
```

So I clone this notebook
Run that code you posted
And the code that renames each json file to what I want?

nono the one i posted is just the silence trimmer