ecker
18 Followers · 0 Following

  • https://git.ecker.tech/ aims to provide a place to share my efforts while maintaining true ownership of my code, as I do not trust GitHub.

    XMR: 4B9TQdkAkBFYrbj5ztvTx89e5LpucPeTSPzemCihdDi9EBnx7btn8RDNZTBz2zihWsjMnDkzn5As1LU6gLv3KQy8BLsZ8SG

  • Joined on 2022-10-10
Repositories
10
 Projects Packages Public Activity Starred Repositories
ecker pushed to master at ecker/ai-voice-cloning 2023-03-22 17:46:38 +00:00
7ba2301341 cleanups, realigning vall-e training
ecker pushed to master at ecker/ai-voice-cloning 2023-03-22 17:45:48 +00:00
aab8637e3b cleanups, realigning vall-e training
ecker commented on issue ecker/ai-voice-cloning#160 2023-03-22 15:37:56 +00:00
Can't train a single good model

It used to be a slider; I forgot I made it a number input because sliders have to have an arbitrary cap and number inputs don't: image

Rega…

ecker pushed to master at ecker/vall-e 2023-03-22 14:44:32 +00:00
fda795bd94 shortcut phonemizing text by using the included phoneme transcriptions when using the prepare librilight script
ecker deleted branch main from ecker/vall-e 2023-03-22 14:15:45 +00:00
ecker commented on issue ecker/ai-voice-cloning#160 2023-03-22 13:12:58 +00:00
Can't train a single good model

I tried redoing it with commit 0231550287 from about 2 weeks ago, and the output was much better; close to the dataset voice. The training ran much faster too.

This repo itself doesn't…

ecker closed issue ecker/ai-voice-cloning#162 2023-03-22 12:48:28 +00:00
ImportError: cannot import name 'do_gc' from 'tortoise.utils.device'
ecker pushed to master at ecker/vall-e 2023-03-22 03:24:31 +00:00
5916d7b2eb modified paths to not need to really define them in the YAML
ecker pushed to master at ecker/vall-e 2023-03-22 03:23:11 +00:00
1fbe80ce6e modified paths to not need to really define them in the YAML
ecker pushed to master at ecker/vall-e 2023-03-22 02:42:41 +00:00
8caf568921 align metric prints a little better with DLAS's
ecker pushed to master at ecker/vall-e 2023-03-21 23:22:23 +00:00
7dd0299fa8 unified exporting of the model (might not even need to bother with exporting)
ecker pushed to master at ecker/ai-voice-cloning 2023-03-21 22:19:04 +00:00
909325bb5a ugh
ecker closed issue ecker/ai-voice-cloning#50 2023-03-21 21:40:56 +00:00
UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 2: character maps to <undefined>
ecker closed issue ecker/ai-voice-cloning#127 2023-03-21 21:40:36 +00:00
Error On Install Dependencies; expected?
ecker closed issue ecker/ai-voice-cloning#137 2023-03-21 21:40:18 +00:00
ModuleNotFoundError: No module named psutil
ecker commented on issue ecker/ai-voice-cloning#156 2023-03-21 21:36:46 +00:00
Possible to use separate batch size for autoregressive samples and CLVP?

Nevermind, forgoing the CLVP just means that you're effectively going to have a batch size = candidate size.

A solution has been implemented in commit 5a5fd9ca87b08d5e8276eca911a164781342b61d. …

ecker closed issue ecker/ai-voice-cloning#156 2023-03-21 21:36:46 +00:00
Possible to use separate batch size for autoregressive samples and CLVP?
ecker pushed to master at ecker/ai-voice-cloning 2023-03-21 21:34:33 +00:00
5a5fd9ca87 Added option to unsqueeze sample batches after sampling
ecker pushed to main at ecker/tortoise-tts 2023-03-21 21:34:00 +00:00
0bcdf81d04 option to decouple sample batch size from CLVP candidate selection size (currently just unsqueezes the batches)
ecker pushed to master at ecker/tortoise-tts 2023-03-21 21:33:56 +00:00
0bcdf81d04 option to decouple sample batch size from CLVP candidate selection size (currently just unsqueezes the batches)