Suggestion: Option to delete intermediate .wav files after processing batches of text #18

New Issue

sakharam_gatne · 2023-02-21T11:24:58Z

sakharam_gatne commented

2023-02-21 11:24:58 +00:00

Firstly, it'll keep the folder structure more organized, as the intermediate files clutter up the results folder, as well as the History tab in Gradio.
Also, it will be better for colab users, as the zipped folder won't take as long to download.

Making it optional will also not disturb users who are comfortable with the current workflow.

Firstly, it'll keep the folder structure more organized, as the intermediate files clutter up the results folder, as well as the History tab in Gradio. Also, it will be better for colab users, as the zipped folder won't take as long to download. Making it optional will also not disturb users who are comfortable with the current workflow.

mrq commented

2023-02-21 13:39:03 +00:00

Great idea. In theory, the pieces remain in case I absolutely need to mix and match generations, but I haven't done that at all yet, and worse case I can just open Tenacity and manually edit there.

It's been a pain seeing a bunch of clutter in the history tab too, but the history tab needs some love as it does some rather abhorrent things for loading.

Great idea. In theory, the pieces remain in case I absolutely need to mix and match generations, but I haven't done that at all yet, and worse case I can just open Tenacity and manually edit there. It's been a pain seeing a bunch of clutter in the history tab too, but the history tab needs some love as it does some rather abhorrent things for loading.

👍 1

sakharam_gatne commented

2023-02-21 14:13:36 +00:00

it does some rather abhorrent things for loading

lol yeah, yesterday I processed one chapter of a book, it generated about 50 audio files, and the history tab froze for a solid minute before loading the list of files.

Also, on a side note, in the colab notebook, if you connect to your gdrive, it creates the (training, voices, results) folders right at the root of the drive. It would probably be better if it created a parent folder for it called "ai-voice-cloning" or something.

> it does some rather abhorrent things for loading lol yeah, yesterday I processed one chapter of a book, it generated about 50 audio files, and the history tab froze for a solid minute before loading the list of files. Also, on a side note, in the colab notebook, if you connect to your gdrive, it creates the (training, voices, results) folders right at the root of the drive. It would probably be better if it created a parent folder for it called "ai-voice-cloning" or something.

mrq commented

2023-02-21 21:53:30 +00:00

Apologies for taking so long to get around to this. A mix of everything interrupting me, and trying to follow through with a possible breakthrough with training on lower VRAM cards.

Added setting Delete Non-Final Output (default on) in 9e64dad785. I also cleaned up some things with handling the post-processing, so the voicefixed output should have its settings embedded (as I was lazy and didn't do this before), and not generate JSONs if embedding to output is on (to avoid useless cluttering, as you can just drop it into the UI to re-grab settings).

Apologies for taking so long to get around to this. A mix of everything interrupting me, and trying to follow through with a possible breakthrough with training on lower VRAM cards. Added setting `Delete Non-Final Output` (default on) in 9e64dad785bc6d6e0e948784b27028d5d4746e25. I also cleaned up some things with handling the post-processing, so the voicefixed output should have its settings embedded (as I was lazy and didn't do this before), and not generate JSONs if embedding to output is on (to avoid useless cluttering, as you can just drop it into the UI to re-grab settings).

👍 1

sakharam_gatne commented

2023-02-22 13:47:44 +00:00

Thanks! Also, any opinion on the second point about the notebook?

mrq commented

2023-02-22 15:23:18 +00:00

Right right, almost forgot. Give me a few to figure out an elegant way to use setups as I'm not well versed in bash scripts.

mrq commented

2023-02-22 15:48:13 +00:00

Added in aa96edde2f. I did a few rudimentary tests (that involved nuking my userdata, but it's not a huge deal since it's an old model) and works fine, no userdata nukes.

Added in aa96edde2fa9072a4a56a89459675a24b655319e. I did a few rudimentary tests (that involved nuking my userdata, but it's not a huge deal since it's an old model) and works fine, no userdata nukes.

sakharam_gatne commented

2023-02-22 16:02:16 +00:00

No problem at all, thanks a lot!

sakharam_gatne closed this issue

2023-02-22 16:02:16 +00:00

Sign in to join this conversation.