MagVITS

VITS with phoneme-level prosody modeling based on MaskGIT （WIP）

feature: inference speed ~= bert-vits2 & prosody > bert-vits2 (maybe)

目前代码正在重构，可能还跑不通，目前不建议跑中文预训练模型不久后会上传（数据：原神中文+aishell 共200h多一些）

Name	Name	Last commit message	Last commit date
Latest commit innnky Aug 31, 2024 41bcaf4 · Aug 31, 2024 History 11 Commits
asr	asr	fix	Aug 11, 2024
configs	configs	fix	Aug 11, 2024
img	img	update readme	Aug 10, 2024
module	module	add transformer flow & update textencoder	Aug 31, 2024
pretrain	pretrain	fix	Aug 11, 2024
text	text	init project	Aug 10, 2024
transformer	transformer	init project	Aug 10, 2024
README.md	README.md	Update README.md	Aug 21, 2024
asr_train.py	asr_train.py	init project	Aug 10, 2024
data_conf.py	data_conf.py	init project	Aug 10, 2024
extract_duration.py	extract_duration.py	fix	Aug 11, 2024
extract_spk_embedding.py	extract_spk_embedding.py	fix	Aug 11, 2024
extract_ssl.py	extract_ssl.py	fix	Aug 11, 2024
gen_filelist.py	gen_filelist.py	fix	Aug 11, 2024
gen_phonemes.py	gen_phonemes.py	fix	Aug 11, 2024
inference.py	inference.py	init project	Aug 10, 2024
requirements.txt	requirements.txt	fix	Aug 11, 2024
resample.py	resample.py	fix	Aug 11, 2024
utils.py	utils.py	update	Aug 10, 2024
vits_train.py	vits_train.py	update	Aug 10, 2024

Provide feedback