gforce commented on issue mrq/ai-voice-cloning#478
Noise at the end of generated voiceCould be temperature or often its just something missed in the training samples.
gforce opened issue mrq/ai-voice-cloning#473
metavoicio / metavoice-1B-v0.1
gforce commented on issue mrq/ai-voice-cloning#314
How to provide Dynamic Prompt Setting Editing (switch between voices)Would love to know how far prompt engineering can go and even the effect punctuation has in the text such as : and other ways of expressing dialogue and context. Seems to be little out there. I…
gforce commented on issue mrq/ai-voice-cloning#253
Results, Retrospectives, and Recommendations> A smart trick I ran into is running generated audio through RVC with its own Harvest trained model, helps smooth out the audio and make it more consistent Have you played around with this?…
gforce commented on issue mrq/ai-voice-cloning#296
Audio artifacts/repetitive words after trainingI wouldn't be surprised if those artifacts and weirdness are because of so many short samples used to train it on. I would also increase the temperature to over .5, and increase diffusion temp…
gforce opened issue mrq/ai-voice-cloning#289
conditioning_length: 44000 is different to sample rate?
gforce commented on issue mrq/ai-voice-cloning#253
Results, Retrospectives, and RecommendationsI started editting it but it was so tedious I decided to redo the sample set again in audacity and actually validate this time to cull the very short segments.
gforce commented on issue mrq/ai-voice-cloning#253
Results, Retrospectives, and RecommendationsHow do you go about manually changing the whisper.json to re-slice without messing with the metrics that the whisper json creates? Will that not cause some sort of conflict or is that extra data…
gforce commented on issue mrq/ai-voice-cloning#253
Results, Retrospectives, and Recommendations> Overall, it's better to be testing the checkpoints along the way and until it no longer changes or begins to get worse. I will disagree with gforce on the autoslicing bit, I feel it's better…
gforce commented on issue mrq/ai-voice-cloning#266
"Original Latents Method" - What is this for, and what is the best option?Its just the method the original repo by neonjib used, I havent made any thorough comparison but I gather the 'new' method only uses the first 4 secs of the sample clip iirc, whereas the old one…
gforce commented on issue mrq/ai-voice-cloning#253
Results, Retrospectives, and RecommendationsHonestly I wing it using the recommended settings more or less and get good results. I've found no considerable difference between 1000 steps and 2500, seems to be an efficient training method. I…
gforce commented on issue mrq/ai-voice-cloning#253
Results, Retrospectives, and RecommendationsYes as usual I have found the quality of the training data is very important, a test model I made had no considerable differences in the output quality whether it was using 12 samples or 50,…
gforce commented on issue mrq/ai-voice-cloning#252
Whats changed recently? finetuning in webUI is completely broken nowSo using that commit above the yaml can be generated and then I just needed a duplicate of each in /training and in subject dataset file, for some reason it changes where it wants the file to be…
gforce commented on issue mrq/ai-voice-cloning#242
Is it possible to generate using the command-line?> Shit, I forgot to do this. Gomen. I'm a bit tied up for the rest of the week, so... > > You ***might*** get lucky with just copying an existing `do_tts.py` (such as from 152334H's fork, I…
gforce commented on issue mrq/ai-voice-cloning#252
Whats changed recently? finetuning in webUI is completely broken nowOh right I see so its the updated google packages that may be stuffing it up, I did revert back to 31da215c5f0f77a249dd5131bf8c3a4f149299a3 and the file pathways worked again enough to actually…
gforce opened issue mrq/ai-voice-cloning#252
Whats changed recently? finetuning in webUI is completely broken now