tuna2134/sbv2-v2
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

init (not checked bat script yet)

Browse Source
This commit is contained in:
litagin02
2023-12-27 05:37:46 +09:00
parent 11a1e7e80d
commit 58fab45b84
235 changed files with 2831 additions and 773509 deletions
Download Patch File Download Diff File Expand all files Collapse all files

242
infer.py
View File

@@ -1,109 +1,35 @@
"""
版本管理、兼容推理及模型加载实现。
版本说明：
1. 版本号与github的release版本号对应，使用哪个release版本训练的模型即对应其版本号
2. 请在模型的config.json中显示声明版本号，添加一个字段"version" : "你的版本号"
特殊版本说明：
1.1.1-fix： 1.1.1版本训练的模型，但是在推理时使用dev的日语修复
2.3：当前版本
"""
import torch
import commons
from text import cleaned_text_to_sequence, get_bert
# from clap_wrapper import get_clap_audio_feature, get_clap_text_feature
from text.cleaner import clean_text
import utils
from models import SynthesizerTrn
from text import cleaned_text_to_sequence, get_bert
from text.cleaner import clean_text
from text.symbols import symbols
from oldVersion.V220.models import SynthesizerTrn as V220SynthesizerTrn
from oldVersion.V220.text import symbols as V220symbols
from oldVersion.V210.models import SynthesizerTrn as V210SynthesizerTrn
from oldVersion.V210.text import symbols as V210symbols
from oldVersion.V200.models import SynthesizerTrn as V200SynthesizerTrn
from oldVersion.V200.text import symbols as V200symbols
from oldVersion.V111.models import SynthesizerTrn as V111SynthesizerTrn
from oldVersion.V111.text import symbols as V111symbols
from oldVersion.V110.models import SynthesizerTrn as V110SynthesizerTrn
from oldVersion.V110.text import symbols as V110symbols
from oldVersion.V101.models import SynthesizerTrn as V101SynthesizerTrn
from oldVersion.V101.text import symbols as V101symbols
from oldVersion import V111, V110, V101, V200, V210, V220
# 当前版本信息
latest_version = "2.3"
# 版本兼容
SynthesizerTrnMap = {
"2.2": V220SynthesizerTrn,
"2.1": V210SynthesizerTrn,
"2.0.2-fix": V200SynthesizerTrn,
"2.0.1": V200SynthesizerTrn,
"2.0": V200SynthesizerTrn,
"1.1.1-fix": V111SynthesizerTrn,
"1.1.1": V111SynthesizerTrn,
"1.1": V110SynthesizerTrn,
"1.1.0": V110SynthesizerTrn,
"1.0.1": V101SynthesizerTrn,
"1.0": V101SynthesizerTrn,
"1.0.0": V101SynthesizerTrn,
}
symbolsMap = {
"2.2": V220symbols,
"2.1": V210symbols,
"2.0.2-fix": V200symbols,
"2.0.1": V200symbols,
"2.0": V200symbols,
"1.1.1-fix": V111symbols,
"1.1.1": V111symbols,
"1.1": V110symbols,
"1.1.0": V110symbols,
"1.0.1": V101symbols,
"1.0": V101symbols,
"1.0.0": V101symbols,
}
# def get_emo_(reference_audio, emotion, sid):
# emo = (
# torch.from_numpy(get_emo(reference_audio))
# if reference_audio and emotion == -1
# else torch.FloatTensor(
# np.load(f"emo_clustering/{sid}/cluster_center_{emotion}.npy")
# )
# )
# return emo
# latest_version = "1.0"
def get_net_g(model_path: str, version: str, device: str, hps):
if version != latest_version:
net_g = SynthesizerTrnMap[version](
len(symbolsMap[version]),
hps.data.filter_length // 2 + 1,
hps.train.segment_size // hps.data.hop_length,
n_speakers=hps.data.n_speakers,
**hps.model,
).to(device)
else:
# 当前版本模型 net_g
net_g = SynthesizerTrn(
len(symbols),
hps.data.filter_length // 2 + 1,
hps.train.segment_size // hps.data.hop_length,
n_speakers=hps.data.n_speakers,
**hps.model,
).to(device)
net_g = SynthesizerTrn(
len(symbols),
hps.data.filter_length // 2 + 1,
hps.train.segment_size // hps.data.hop_length,
n_speakers=hps.data.n_speakers,
**hps.model,
).to(device)
net_g.state_dict()
_ = net_g.eval()
_ = utils.load_checkpoint(model_path, net_g, None, skip_optimizer=True)
if model_path.endswith(".pth") or model_path.endswith(".pt"):
_ = utils.load_checkpoint(model_path, net_g, None, skip_optimizer=True)
elif model_path.endswith(".safetensors"):
_ = utils.load_safetensors(model_path, net_g, device)
else:
raise ValueError(f"Unknown model format: {model_path}")
return net_g
def get_text(text, language_str, hps, device, style_text=None, style_weight=0.7):
style_text = None if style_text == "" else style_text
# 在此处实现当前版本的get_text
norm_text, phone, tone, word2ph = clean_text(text, language_str)
phone, tone, language = cleaned_text_to_sequence(phone, tone, language_str)
@@ -123,15 +49,15 @@ def get_text(text, language_str, hps, device, style_text=None, style_weight=0.7)
if language_str == "ZH":
bert = bert_ori
ja_bert = torch.randn(1024, len(phone))
en_bert = torch.randn(1024, len(phone))
ja_bert = torch.zeros(1024, len(phone))
en_bert = torch.zeros(1024, len(phone))
elif language_str == "JP":
bert = torch.randn(1024, len(phone))
bert = torch.zeros(1024, len(phone))
ja_bert = bert_ori
en_bert = torch.randn(1024, len(phone))
en_bert = torch.zeros(1024, len(phone))
elif language_str == "EN":
bert = torch.randn(1024, len(phone))
ja_bert = torch.randn(1024, len(phone))
bert = torch.zeros(1024, len(phone))
ja_bert = torch.zeros(1024, len(phone))
en_bert = bert_ori
else:
raise ValueError("language_str should be ZH, JP or EN")
@@ -148,123 +74,21 @@ def get_text(text, language_str, hps, device, style_text=None, style_weight=0.7)
def infer(
text,
emotion,
style_vec,
sdp_ratio,
noise_scale,
noise_scale_w,
length_scale,
sid,
sid: int, # In the original Bert-VITS2, its speaker_name: str, but here it's id
language,
hps,
net_g,
device,
reference_audio=None,
skip_start=False,
skip_end=False,
style_text=None,
style_weight=0.7,
):
# 2.2版本参数位置变了
inferMap_V4 = {
"2.2": V220.infer,
}
# 2.1 参数新增 emotion reference_audio skip_start skip_end
inferMap_V3 = {
"2.1": V210.infer,
}
# 支持中日英三语版本
inferMap_V2 = {
"2.0.2-fix": V200.infer,
"2.0.1": V200.infer,
"2.0": V200.infer,
"1.1.1-fix": V111.infer_fix,
"1.1.1": V111.infer,
"1.1": V110.infer,
"1.1.0": V110.infer,
}
# 仅支持中文版本
# 在测试中，并未发现两个版本的模型不能互相通用
inferMap_V1 = {
"1.0.1": V101.infer,
"1.0": V101.infer,
"1.0.0": V101.infer,
}
version = hps.version if hasattr(hps, "version") else latest_version
# 非当前版本，根据版本号选择合适的infer
if version != latest_version:
if version in inferMap_V4.keys():
emotion = "" # Use empty emotion prompt
return inferMap_V4[version](
text,
emotion,
sdp_ratio,
noise_scale,
noise_scale_w,
length_scale,
sid,
language,
hps,
net_g,
device,
reference_audio,
skip_start,
skip_end,
style_text,
style_weight,
)
if version in inferMap_V3.keys():
emotion = 0
return inferMap_V3[version](
text,
sdp_ratio,
noise_scale,
noise_scale_w,
length_scale,
sid,
language,
hps,
net_g,
device,
reference_audio,
emotion,
skip_start,
skip_end,
style_text,
style_weight,
)
if version in inferMap_V2.keys():
return inferMap_V2[version](
text,
sdp_ratio,
noise_scale,
noise_scale_w,
length_scale,
sid,
language,
hps,
net_g,
device,
)
if version in inferMap_V1.keys():
return inferMap_V1[version](
text,
sdp_ratio,
noise_scale,
noise_scale_w,
length_scale,
sid,
hps,
net_g,
device,
)
# 在此处实现当前版本的推理
# emo = get_emo_(reference_audio, emotion, sid)
# if isinstance(reference_audio, np.ndarray):
# emo = get_clap_audio_feature(reference_audio, device)
# else:
# emo = get_clap_text_feature(emotion, device)
# emo = torch.squeeze(emo, dim=1)
bert, ja_bert, en_bert, phones, tones, lang_ids = get_text(
text,
language,
@@ -295,19 +119,20 @@ def infer(
ja_bert = ja_bert.to(device).unsqueeze(0)
en_bert = en_bert.to(device).unsqueeze(0)
x_tst_lengths = torch.LongTensor([phones.size(0)]).to(device)
# emo = emo.to(device).unsqueeze(0)
style_vec = torch.from_numpy(style_vec).to(device).unsqueeze(0)
del phones
speakers = torch.LongTensor([hps.data.spk2id[sid]]).to(device)
sid_tensor = torch.LongTensor([sid]).to(device)
audio = (
net_g.infer(
x_tst,
x_tst_lengths,
speakers,
sid_tensor,
tones,
lang_ids,
bert,
ja_bert,
en_bert,
style_vec=style_vec,
sdp_ratio=sdp_ratio,
noise_scale=noise_scale,
noise_scale_w=noise_scale_w,
@@ -323,9 +148,10 @@ def infer(
lang_ids,
bert,
x_tst_lengths,
speakers,
sid_tensor,
ja_bert,
en_bert,
style_vec,
) # , emo
if torch.cuda.is_available():
torch.cuda.empty_cache()
@@ -334,6 +160,7 @@ def infer(
def infer_multilang(
text,
style_vec,
sdp_ratio,
noise_scale,
noise_scale_w,
@@ -343,8 +170,6 @@ def infer_multilang(
hps,
net_g,
device,
reference_audio=None,
emotion=None,
skip_start=False,
skip_end=False,
):
@@ -413,6 +238,7 @@ def infer_multilang(
bert,
ja_bert,
en_bert,
style_vec=style_vec,
sdp_ratio=sdp_ratio,
noise_scale=noise_scale,
noise_scale_w=noise_scale_w,