diff --git a/README.md b/README.md
index 342c422..43c4d96 100644
--- a/README.md
+++ b/README.md
@@ -40,12 +40,10 @@ We plan to support a variaty of open source TTS datasets, include but not limite
 
 ## Pretrained Models
 
-| Dataset | Language | Checkpoint Model | Runtime Model |
-| ------- | -------- | ---------------- | ------------- |
-| Baker   | CN       | [BERT](https://wenet.org.cn/downloads?models=wetts&version=baker_bert_exp.tar.gz) | [BERT](https://wenet.org.cn/downloads?models=wetts&version=baker_bert_onnx.tar.gz) |
-| Baker   | CN       | [VITS](https://wenet.org.cn/downloads?models=wetts&version=baker_vits_v1_exp.tar.gz) | [VITS](https://wenet.org.cn/downloads?models=wetts&version=baker_vits_v1_onnx.tar.gz) |
-
-English G2P model: [english_us_arpa v2.0.0a](https://wenet.org.cn/downloads?models=wetts&version=g2p_en.tar.gz), powered by [MFA](https://github.com/MontrealCorpusTools/mfa-models/releases/tag/g2p-english_us_arpa-v2.0.0a).
+| Dataset        | Language | Checkpoint Model | Runtime Model |
+| -------------- | -------- | ---------------- | ------------- |
+| Baker          | CN       | [BERT](https://wenet.org.cn/downloads?models=wetts&version=baker_bert_exp.tar.gz) | [BERT](https://wenet.org.cn/downloads?models=wetts&version=baker_bert_onnx.tar.gz) |
+| Multilingual   | CN       | [VITS](https://wenet.org.cn/downloads?models=wetts&version=multilingual_vits_v3_exp.tar.gz) | [VITS](https://wenet.org.cn/downloads?models=wetts&version=multilingual_vits_v3_onnx.tar.gz) |
 
 ## Runtime
 
@@ -64,21 +62,10 @@ cd runtime/onnxruntime
 cmake -B build -DCMAKE_BUILD_TYPE=Release
 cmake --build build
 ./build/bin/tts_main \
-  --tagger baker_bert_onnx/zh_tn_tagger.fst \
-  --verbalizer baker_bert_onnx/zh_tn_verbalizer.fst \
-  --vocab baker_bert_onnx/vocab.txt \
-  --char2pinyin baker_bert_onnx/pinyin_dict.txt \
-  --pinyin2id baker_bert_onnx/polyphone_phone.txt \
-  --pinyin2phones baker_bert_onnx/lexicon.txt \
-  --g2p_prosody_model baker_bert_onnx/19.onnx \
-  --speaker2id baker_vits_v1_onnx/speaker.txt \
+  --frontend_flags baker_bert_onnx/frontend.flags \
+  --vits_flags multilingual_vits_v3_onnx/vits.flags \
   --sname baker \
-  --phone2id baker_vits_v1_onnx/phones.txt \
-  --vits_model baker_vits_v1_onnx/G_250000.onnx \
-  --text "你好，我是小明。" \
-  --cmudict g2p_en/cmudict.dict \  # optional
-  --g2p_en_model g2p_en/model.fst \  # optional
-  --g2p_en_sym g2p_en/phones.sym \  # optional
+  --text "hello我是小明。" \
   --wav_path audio.wav
 ```
 
diff --git a/runtime/core/bin/tts_main.cc b/runtime/core/bin/tts_main.cc
index 01495b9..7a5f1fe 100644
--- a/runtime/core/bin/tts_main.cc
+++ b/runtime/core/bin/tts_main.cc
@@ -22,6 +22,10 @@
 #include "model/tts_model.h"
 #include "utils/string.h"
 
+// Flags
+DEFINE_string(frontend_flags, "", "frontend flags file");
+DEFINE_string(vits_flags, "", "vits flags file");
+
 // Text Normalization
 DEFINE_string(tagger, "", "tagger fst file");
 DEFINE_string(verbalizer, "", "verbalizer fst file");
@@ -43,16 +47,18 @@ DEFINE_string(g2p_prosody_model, "", "g2p prosody model file");
 // VITS
 DEFINE_string(speaker2id, "", "speaker to id");
 DEFINE_string(phone2id, "", "phone to id");
-DEFINE_string(sname, "", "speaker name");
 DEFINE_string(vits_model, "", "e2e tts model file");
-
 DEFINE_int32(sampling_rate, 22050, "sampling rate of pcm");
+
+DEFINE_string(sname, "", "speaker name");
 DEFINE_string(text, "", "input text");
 DEFINE_string(wav_path, "", "output wave path");
 
 int main(int argc, char* argv[]) {
   gflags::ParseCommandLineFlags(&argc, &argv, false);
   google::InitGoogleLogging(argv[0]);
+  gflags::ReadFromFlagsFile(FLAGS_frontend_flags, "", false);
+  gflags::ReadFromFlagsFile(FLAGS_vits_flags, "", false);
 
   auto tn = std::make_shared<wetext::Processor>(FLAGS_tagger, FLAGS_verbalizer);
 
diff --git a/wetts/vits/inference_onnx.py b/wetts/vits/inference_onnx.py
index 1913dff..9adeea6 100644
--- a/wetts/vits/inference_onnx.py
+++ b/wetts/vits/inference_onnx.py
@@ -62,7 +62,8 @@ def main():
     scales = scales.unsqueeze(0)
 
     for line in open(args.test_file):
-        audio_path, sid, text = line.strip().split("|")
+        audio_path, speaker, text = line.strip().split("|")
+        sid = speaker_dict[speaker]
         seq = [phone_dict[symbol] for symbol in text.split()]
 
         x = torch.LongTensor([seq])
diff --git a/wetts/vits/utils.py b/wetts/vits/utils.py
index d433eef..774008f 100644
--- a/wetts/vits/utils.py
+++ b/wetts/vits/utils.py
@@ -4,8 +4,6 @@
 import logging
 import os
 
-import numpy as np
-from scipy.io.wavfile import read
 import torch
 
 MATPLOTLIB_FLAG = False
@@ -152,11 +150,6 @@ def plot_alignment_to_numpy(alignment, info=None):
     return data
 
 
-def load_wav_to_torch(full_path):
-    sampling_rate, data = read(full_path)
-    return torch.FloatTensor(data.astype(np.float32)), sampling_rate
-
-
 def load_filepaths_and_text(filename, split="|"):
     with open(filename, encoding="utf-8") as f:
         filepaths_and_text = [line.strip().split(split) for line in f]
@@ -217,8 +210,7 @@ def get_hparams(init=True):
 def get_hparams_from_dir(model_dir):
     config_save_path = os.path.join(model_dir, "config.json")
     with open(config_save_path, "r") as f:
-        data = f.read()
-    config = json.loads(data)
+        config = json.load(f)
 
     hparams = HParams(**config)
     hparams.model_dir = model_dir
@@ -227,8 +219,7 @@ def get_hparams_from_dir(model_dir):
 
 def get_hparams_from_file(config_path):
     with open(config_path, "r") as f:
-        data = f.read()
-    config = json.loads(data)
+        config = json.load(f)
 
     hparams = HParams(**config)
     return hparams