Bro-Xun / MoeGoe-generating-workflow

Executable file for VITS inference

MoeGoe generating-process workflow

origined from CjangCjengh/MoeGoe

usage

• install requirements in requirements.txt
• prepare the argument list (if choosing t mode)
• edit config.yml
• run MoeGoe.py

to prepare argument list

---

a simple example:

#会员制餐厅 [4]
#对话
本餐厅就是为了那些对各种美味食物已经感到厌倦了的人们 [1]
为他们提供与之身份相应的食物
 [2]
欢迎光临~
欢迎光临~ 2 [3]

---

[1]: use a whole line to express one sentence

[2]: you are able to use a blank line to make the list easier to read

[3]: use [blank]2(a number) to choose the model this sentence uses (default [blank]1)

[4]: use # at the start of a line to create an annotation line (dont use [3] [4] together)

it must be a plain text file.

about config.yml

#please make sure the number of your models before you submit
config_path: (your full path of config file)
model_path: (your full path of model file)
##########
## choose one of following options
argument_path: (your full path of argument list)
argument_lang: (the language you want to choose) #pay attention to supported languages of your model!!!
##[ZH] [JA] [KO] [SA] [EN] avaliable
#OR
audio_dir: ###audio_directory: audio input must be in form of ms-wav(*.wav)
origin: ###Original speaker ID
target: ###Target speaker ID
##########
output_path: F:\Desktop\haochi\ #(full path)
method: #t for text2voice while v for voice2voice
time_gap: #time gap between each process (seconds)

useful links

• Pretrained models

thanks the original developer

---

_{feel free to pull PR or write issues}