|
--- |
|
license: other |
|
--- |
|
|
|
https://vocaroo.com/166irX2cY4g0 |
|
|
|
Nyanners narrator for bark and RVC. |
|
|
|
Use with something like https://github.com/gitmylo/audio-webui |
|
|
|
------------------------------------------------------------------------ |
|
|
|
Added index files for better inference. Unzip them into the folder with the model. |
|
They're large but improve things quite a bit. Also a new model, soon to be 2 that |
|
was done on cleaned datasets and trained for much longer. |
|
|
|
It is now able to sing decently but still won't transform a voice from male. |
|
|
|
|
|
|
|
------------------------------------------------------------------------ |
|
One model is done on a single video. The other is done on multiple similar videos. Something like 2hrs vs 6hrs of audio. |
|
|
|
They are sorted in terms of steps or epochs that were trained. Longer isn't always better except for some samples. |
|
|
|
Generate a voice with the bark NPZ for best effect but it can definitely work on other decent sounding TTS output. |
|
|
|
Have it be similar to the output you want. It won't necesarily do well with a male TTS voice. |
|
|
|
Music can be done but I think an SVC model will be better. |
