• 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
ecker pushed to master at ecker/ai-voice-cloning 2023-03-11 17:03:00 +00:00
e3fdb79b49 rocm5.2 works for me desu so I bumped it back up
ecker pushed to master at ecker/ai-voice-cloning 2023-03-11 16:55:39 +00:00
e680d84a13 removed the hotfix pip installs that whisperx requires now that whisperx is gone
ecker commented on issue ecker/ai-voice-cloning#101 2023-03-11 16:47:19 +00:00
Stack expects a non-empty TensorList

Ah, I managed to have it happen myself. Should be remedied in commit cf41492f76d2bb93cb2972bb5585edcf10d531ca. If not:

  • at least for me, it cropped up from having a dataset prepared, but didn't…
ecker closed issue ecker/ai-voice-cloning#101 2023-03-11 16:47:19 +00:00
Stack expects a non-empty TensorList
ecker pushed to master at ecker/ai-voice-cloning 2023-03-11 16:46:07 +00:00
cf41492f76 fall back to normal behavior if theres actually no audiofiles loaded from the dataset when using it for computing latents
ecker closed issue ecker/ai-voice-cloning#109 2023-03-11 16:43:28 +00:00
Cannot reshape tensor / Training process hangs instead of closing
ecker pushed to master at ecker/ai-voice-cloning 2023-03-11 16:40:45 +00:00
b90c164778 Farewell, parasite
ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 16:36:12 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

Pushed commit 2424c455cb9614003c072f6cdc25fa80ba2694ba. It seems every passing day I regret more and more adding whisperx.

I'm very, very tempted to just remove it. It caused nothing but trouble.

ecker pushed to master at ecker/ai-voice-cloning 2023-03-11 16:32:45 +00:00
2424c455cb added option to not slice audio when transcribing, added option to prepare validation dataset on audio duration, added a warning if youre using whisperx and you're slicing audio
ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 16:22:16 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

Well shit.

I just transcribed some more datasets with whisperx+large-v2 and they're consistently cut off too soon at the end. I compared them against whisper and whisper does a better job at…

ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 04:38:42 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

Yeh I agree, though that's why I mentioned it would be nice if this was able to be automated... as I'll have to manually remove ~150 entries from the text file lol.

Just run the `Prepare…

ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 04:37:33 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

But there is also a lot (158 out of 1343 files) which are 0s. Some still have a single word or two in there, but there's a bunch which seem to be completely empty (they probably have a…

ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 03:35:38 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

It should be good, my main issue with whisper is that I'm seeing a lot of single words that get segmented off. I just need to also evaluate how intrusive it'll be to implement it.

ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 03:32:19 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

Say I take my 1 minute source, and cut it into sentences. Can whisper then take each sentence and not cut it further? That's what comes to mind when thinking how to stop whisper from cuttin off…

ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 03:31:22 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

Oh, no, that file is in the voices/patrick folder. In the training/patrick/audio folder its been cut up by whisper into a bunch of short files.

Ah, then it shouldn't affect it, as the…

ecker commented on pull request ecker/ai-voice-cloning#112 2023-03-11 03:28:38 +00:00
master

Oh neat, I can just squash it instead.

ecker pushed to master at ecker/ai-voice-cloning 2023-03-11 03:28:09 +00:00
dcdcf8516c master (#112)
ecker merged pull request ecker/ai-voice-cloning#112 2023-03-11 03:28:05 +00:00
master
ecker commented on pull request ecker/ai-voice-cloning#112 2023-03-11 03:26:12 +00:00
master

Oh right, I forgot to look at the PR. I honestly might just snip out the change so it doesn't have the delete-then-add a whole file thing.

./venv/bin/activate: line 3: $'\r': command not…

ecker commented on issue ecker/ai-voice-cloning#113 2023-03-11 03:21:10 +00:00
Generated voices from training data always garbled.... but works fine using tortoise-tts-fast ... (?)

ahhhh... I dooooo have a sound file which is 1hour+ because it's a narrated audiobook file (with the bad bits snipped out, but still kept as one single file).

That'll do it. If it's in…