diff --git a/examples/aishell3/tts3/README.md b/examples/aishell3/tts3/README.md
index 82b69ad84..fe4887b93 100644
--- a/examples/aishell3/tts3/README.md
+++ b/examples/aishell3/tts3/README.md
@@ -24,7 +24,7 @@ Assume the path to the dataset is `~/datasets/data_aishell3`.
 Assume the path to the MFA result of AISHELL-3 is `./aishell3_alignment_tone`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
diff --git a/examples/aishell3/vc0/README.md b/examples/aishell3/vc0/README.md
index 28fea6292..2f1b37ee2 100644
--- a/examples/aishell3/vc0/README.md
+++ b/examples/aishell3/vc0/README.md
@@ -1,7 +1,7 @@
 # Tacotron2 + AISHELL-3 Voice Cloning
 This example contains code used to train a [Tacotron2 ](https://arxiv.org/abs/1712.05884) model with [AISHELL-3](http://www.aishelltech.com/aishell_3). The trained model can be used in Voice Cloning Task, We refer to the model structure of  [Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis](https://arxiv.org/pdf/1806.04558.pdf) . The general steps are as follows:
 1. Speaker Encoder: We  use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in Tacotron2, because the  transcriptions are not needed, we use more datasets, refer to  [ge2e](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/ge2e).
-2. Synthesizer: Then, we use the trained speaker encoder to generate utterance embedding for each  sentence in AISHELL-3. This embedding is a extra input of  Tacotron2 which will be concated with encoder outputs.
+2. Synthesizer: We use the trained speaker encoder to generate speaker embedding for each sentence in AISHELL-3. This embedding is an extra input of  Tacotron2 which will be concated with encoder outputs.
 3. Vocoder: We use WaveFlow as the neural Vocoder, refer to [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0).
 
 ## Get Started
@@ -10,7 +10,7 @@ Assume the path to the MFA result of AISHELL-3 is `./alignment`.
 Assume the path to the pretrained ge2e model is `ge2e_ckpt_path=./ge2e_ckpt_0.3/step-3000000`
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. start a voice cloning inference.
 ```bash
@@ -20,8 +20,8 @@ Run the command below to
 ```bash
 CUDA_VISIBLE_DEVICES=${gpus} ./local/preprocess.sh ${input} ${preprocess_path} ${alignment} ${ge2e_ckpt_path}
 ```
-#### generate utterance embedding
- Use pretrained GE2E (speaker encoder) to generate utterance embedding for each sentence in AISHELL-3, which has the same file structure with wav files and the format is  `.npy`.
+#### generate speaker embedding
+ Use pretrained GE2E (speaker encoder) to generate speaker embedding for each sentence in AISHELL-3, which has the same file structure with wav files and the format is  `.npy`.
 
 ```bash
 if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
@@ -33,8 +33,8 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
 fi
 ```
 
-The computing time of  utterance embedding can be x hours.
-####  process wav
+The computing time of utterance embedding can be x hours.
+#### process wav
 There are silence in the edge of AISHELL-3's wavs, and the audio amplitude is very small, so, we need to remove the silence and normalize the audio. You can the silence remove method based on   volume or energy, but the effect is not very good, We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get  the alignment of text and  speech, then utilize the alignment results to remove the silence.
 
 We use Montreal Force Aligner 1.0. The label in  aishell3 include pinyin，so the lexicon we provided to MFA is pinyin rather than Chinese characters. And the prosody marks(`$`  and `%`) need to be removed. You shoud preprocess the dataset into the format  which MFA needs, the texts have the same name with wavs and have the suffix `.lab`.
@@ -73,7 +73,7 @@ if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
 fi
 ```
 
-###  Train the model
+### Train the model
 ```bash
 CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${preprocess_path} ${train_output_path}
 ```
@@ -81,7 +81,7 @@ CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${preprocess_path} ${train_output_
 Our model remve  stop token prediction in Tacotron2, because of the problem of extremely unbalanced proportion of positive and negative samples of stop token prediction, and it's very sensitive to the clip of audio silence. We use the last symbol from the highest point of attention to the encoder side as the termination condition.
 
 In addition, in order to accelerate the convergence of the model, we add `guided attention loss` to induce the alignment between encoder and decoder to show diagonal lines faster.
-###  Infernece
+### Voice Cloning
 ```bash
 CUDA_VISIBLE_DEVICES=${gpus} ./local/voice_cloning.sh ${ge2e_params_path} ${tacotron2_params_path} ${waveflow_params_path} ${vc_input} ${vc_output}
 ```
diff --git a/examples/aishell3/vc1/README.md b/examples/aishell3/vc1/README.md
index 8c0aec3af..834942fa0 100644
--- a/examples/aishell3/vc1/README.md
+++ b/examples/aishell3/vc1/README.md
@@ -1,89 +1,129 @@
 # FastSpeech2 + AISHELL-3 Voice Cloning
-This example contains code used to train a [Tacotron2 ](https://arxiv.org/abs/1712.05884) model with [AISHELL-3](http://www.aishelltech.com/aishell_3). The trained model can be used in Voice Cloning Task, We refer to the model structure of  [Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis](https://arxiv.org/pdf/1806.04558.pdf) . The general steps are as follows:
-1. Speaker Encoder: We  use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in Tacotron2, because the  transcriptions are not needed, we use more datasets, refer to  [ge2e](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/ge2e).
-2. Synthesizer: Then, we use the trained speaker encoder to generate utterance embedding for each  sentence in AISHELL-3. This embedding is a extra input of  Tacotron2 which will be concated with encoder outputs.
-3. Vocoder: We use WaveFlow as the neural Vocoder, refer to [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0).
+This example contains code used to train a [FastSpeech2](https://arxiv.org/abs/2006.04558) model with [AISHELL-3](http://www.aishelltech.com/aishell_3). The trained model can be used in Voice Cloning Task, We refer to the model structure of  [Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis](https://arxiv.org/pdf/1806.04558.pdf) . The general steps are as follows:
+1. Speaker Encoder: We  use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in `FastSpeech2`, because the  transcriptions are not needed, we use more datasets, refer to  [ge2e](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/ge2e).
+2. Synthesizer: We use the trained speaker encoder to generate speaker embedding for each sentence in AISHELL-3. This embedding is an extra input of  `FastSpeech2` which will be concated with encoder outputs.
+3. Vocoder: We use [Parallel Wave GAN](http://arxiv.org/abs/1910.11480) as the neural Vocoder, refer to [voc1](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1).
+
+## Dataset
+### Download and Extract the datasaet
+Download AISHELL-3.
+```bash
+wget https://www.openslr.org/resources/93/data_aishell3.tgz
+```
+Extract AISHELL-3.
+```bash
+mkdir data_aishell3
+tar zxvf data_aishell3.tgz -C data_aishell3
+```
+### Get MFA result of AISHELL-3 and Extract it
+We use [MFA2.x](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for aishell3_fastspeech2.
+You can download from here [aishell3_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/AISHELL-3/with_tone/aishell3_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
+
+## Pretrained GE2E model
+We use pretrained GE2E model to generate spwaker embedding for each sentence.
+
+Download pretrained GE2E model from here [ge2e_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/ge2e_ckpt_0.3.zip), and `unzip` it.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/data_aishell3`.
-Assume the path to the MFA result of AISHELL-3 is `./alignment`.
-Assume the path to the pretrained ge2e model is `ge2e_ckpt_path=./ge2e_ckpt_0.3/step-3000000`
+Assume the path to the MFA result of AISHELL-3 is `./aishell3_alignment_tone`.
+Assume the path to the pretrained ge2e model is `./ge2e_ckpt_0.3`.
+
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
-4. start a voice cloning inference.
+4. synthesize waveform from `metadata.jsonl`.
+5. start a voice cloning inference.
 ```bash
 ./run.sh
 ```
 ### Preprocess the dataset
 ```bash
-CUDA_VISIBLE_DEVICES=${gpus} ./local/preprocess.sh ${input} ${preprocess_path} ${alignment} ${ge2e_ckpt_path}
+CUDA_VISIBLE_DEVICES=${gpus} ./local/preprocess.sh ${conf_path} ${ge2e_ckpt_path}
 ```
-#### generate utterance embedding
- Use pretrained GE2E (speaker encoder) to generate utterance embedding for each sentence in AISHELL-3, which has the same file structure with wav files and the format is  `.npy`.
-
-```bash
-if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
-    python3 ${BIN_DIR}/../ge2e/inference.py \
-        --input=${input} \
-        --output=${preprocess_path}/embed \
-        --ngpu=1 \
-        --checkpoint_path=${ge2e_ckpt_path}
-fi
+When it is done. A `dump` folder is created in the current directory. The structure of the dump folder is listed below.
+```text
+dump
+├── dev
+│   ├── norm
+│   └── raw
+├── embed
+│   ├── SSB0005
+│   ├── SSB0009
+│   ├── ...
+│   └── ...
+├── phone_id_map.txt
+├── speaker_id_map.txt
+├── test
+│   ├── norm
+│   └──  raw
+└── train
+    ├── energy_stats.npy
+    ├── norm
+    ├── pitch_stats.npy
+    ├── raw
+    └── speech_stats.npy
 ```
+The `embed` contains the generated speaker embedding for each sentence in AISHELL-3, which has the same file structure with wav files and the format is  `.npy`.
 
 The computing time of  utterance embedding can be x hours.
-####  process wav
-There are silence in the edge of AISHELL-3's wavs, and the audio amplitude is very small, so, we need to remove the silence and normalize the audio. You can the silence remove method based on   volume or energy, but the effect is not very good, We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get  the alignment of text and  speech, then utilize the alignment results to remove the silence.
 
-We use Montreal Force Aligner 1.0. The label in  aishell3 include pinyin，so the lexicon we provided to MFA is pinyin rather than Chinese characters. And the prosody marks(`$`  and `%`) need to be removed. You shoud preprocess the dataset into the format  which MFA needs, the texts have the same name with wavs and have the suffix `.lab`.
+The dataset is split into 3 parts, namely `train`, `dev` and` test`, each of which contains a `norm` and `raw` sub folder. The raw folder contains speech、pitch and energy features of each utterances, while the norm folder contains normalized ones. The statistics used to normalize features are computed from the training set, which is located in `dump/train/*_stats.npy`.
 
-We use [lexicon.txt](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/paddlespeech/t2s/exps/voice_cloning/tacotron2_ge2e/lexicon.txt) as the lexicon.
+Also there is a `metadata.jsonl` in each subfolder. It is a table-like file which contains phones, text_lengths, speech_lengths, durations, path of speech features, path of pitch features, path of energy features, speaker and id of each utterance.
 
-You can download the alignment results from here [alignment_aishell3.tar.gz](https://paddlespeech.bj.bcebos.com/Parakeet/alignment_aishell3.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
+The preprocessing step is very similar to that one of [tts3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/tts3), but  there is one more `ge2e/inference` step here.
 
+###  Train the model
+`./local/train.sh` calls `${BIN_DIR}/train.py`.
 ```bash
-if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
-    echo "Process wav ..."
-    python3 ${BIN_DIR}/process_wav.py \
-        --input=${input}/wav \
-        --output=${preprocess_path}/normalized_wav \
-        --alignment=${alignment}
-fi
+CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${conf_path} ${train_output_path}
 ```
+The training step is very similar to that one of [tts3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/tts3), but  we should set `--voice-cloning=True` when calling `${BIN_DIR}/train.py`.
 
-#### preprocess transcription
-We revert the transcription into `phones` and  `tones`. It is worth noting that our processing here is different from that used for MFA, we separated the tones. This is a processing method, of course, you can only segment initials and vowels.
-
+### Synthesize
+We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1) as the neural vocoder.
+Download pretrained parallel wavegan model from [pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_aishell3_ckpt_0.5.zip) and unzip it.
 ```bash
-if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
-    python3 ${BIN_DIR}/preprocess_transcription.py \
-        --input=${input} \
-        --output=${preprocess_path}
-fi
+unzip pwg_aishell3_ckpt_0.5.zip
 ```
-The default input is  `~/datasets/data_aishell3/train`，which contains `label_train-set.txt`, the processed results are `metadata.yaml` and  `metadata.pickle`. the former is a text format for easy viewing, and the latter is a binary format for direct reading.
-#### extract mel
-```python
-if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
-    python3 ${BIN_DIR}/extract_mel.py \
-        --input=${preprocess_path}/normalized_wav \
-        --output=${preprocess_path}/mel
-fi
+Parallel WaveGAN checkpoint contains files listed below.
+```text
+pwg_aishell3_ckpt_0.5
+├── default.yaml                   # default config used to train parallel wavegan
+├── feats_stats.npy                # statistics used to normalize spectrogram when training parallel wavegan
+└── snapshot_iter_1000000.pdz      # generator parameters of parallel wavegan
 ```
-
-###  Train the model
+`./local/synthesize.sh` calls `${BIN_DIR}/synthesize.py`, which can synthesize waveform from `metadata.jsonl`.
 ```bash
-CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${preprocess_path} ${train_output_path}
+CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name}
 ```
+The synthesizing step is very similar to that one of [tts3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/tts3), but  we should set `--voice-cloning=True` when calling `${BIN_DIR}/synthesize.py`.
 
-Our model remve  stop token prediction in Tacotron2, because of the problem of extremely unbalanced proportion of positive and negative samples of stop token prediction, and it's very sensitive to the clip of audio silence. We use the last symbol from the highest point of attention to the encoder side as the termination condition.
+### Voice Cloning
+Assume there are some  reference audios in `./ref_audio`
+```text
+ref_audio
+├── 001238.wav
+├── LJ015-0254.wav
+└── audio_self_test.mp3
+```
+`./local/voice_cloning.sh` calls `${BIN_DIR}/voice_cloning.py`
 
-In addition, in order to accelerate the convergence of the model, we add `guided attention loss` to induce the alignment between encoder and decoder to show diagonal lines faster.
-###  Infernece
 ```bash
-CUDA_VISIBLE_DEVICES=${gpus} ./local/voice_cloning.sh ${ge2e_params_path} ${tacotron2_params_path} ${waveflow_params_path} ${vc_input} ${vc_output}
+CUDA_VISIBLE_DEVICES=${gpus} ./local/voice_cloning.sh ${conf_path} ${train_output_path} ${ckpt_name} ${ge2e_params_path} ${ref_audio_dir}
 ```
 ## Pretrained Model
-[tacotron2_aishell3_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/tacotron2_aishell3_ckpt_0.3.zip).
+[fastspeech2_nosil_aishell3_vc1_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_aishell3_vc1_ckpt_0.5.zip)
+
+FastSpeech2 checkpoint contains files listed below.
+(There is no need for `speaker_id_map.txt` here )
+
+```text
+fastspeech2_nosil_aishell3_ckpt_vc1_0.5
+├── default.yaml            # default config used to train fastspeech2
+├── phone_id_map.txt        # phone vocabulary file when training fastspeech2
+├── snapshot_iter_96400.pdz # model parameters and optimizer states
+└── speech_stats.npy        # statistics used to normalize spectrogram when training fastspeech2
+```
diff --git a/examples/aishell3/voc1/README.md b/examples/aishell3/voc1/README.md
index d9e8ce591..d67af726e 100644
--- a/examples/aishell3/voc1/README.md
+++ b/examples/aishell3/voc1/README.md
@@ -22,7 +22,7 @@ Assume the path to the dataset is `~/datasets/data_aishell3`.
 Assume the path to the MFA result of AISHELL-3 is `./aishell3_alignment_tone`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
diff --git a/examples/csmsc/tts2/README.md b/examples/csmsc/tts2/README.md
index 2088ed15c..61c4972bc 100644
--- a/examples/csmsc/tts2/README.md
+++ b/examples/csmsc/tts2/README.md
@@ -14,7 +14,7 @@ Assume the path to the dataset is `~/datasets/BZNSYP`.
 Assume the path to the MFA result of CSMSC is `./baker_alignment_tone`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
@@ -208,7 +208,8 @@ CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path}
 ```
 
 ## Pretrained Model
-Pretrained SpeedySpeech model with no silence in the edge of audios. [speedyspeech_nosil_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_ckpt_0.5.zip)
+Pretrained SpeedySpeech model with no silence in the edge of audios[speedyspeech_nosil_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_ckpt_0.5.zip).
+
 Static model can be downloaded here [speedyspeech_nosil_baker_static_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_static_0.5.zip).
 
 SpeedySpeech checkpoint contains files listed below.
diff --git a/examples/csmsc/tts3/README.md b/examples/csmsc/tts3/README.md
index 6e4701dfe..6570d33dc 100644
--- a/examples/csmsc/tts3/README.md
+++ b/examples/csmsc/tts3/README.md
@@ -14,7 +14,7 @@ Assume the path to the dataset is `~/datasets/BZNSYP`.
 Assume the path to the MFA result of CSMSC is `./baker_alignment_tone`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
@@ -199,8 +199,9 @@ CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path}
 ```
 
 ## Pretrained Model
-Pretrained FastSpeech2 model with no silence in the edge of audios. [fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip)
-Static model can be downloaded here [fastspeech2_nosil_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_static_0.4.zip)
+Pretrained FastSpeech2 model with no silence in the edge of audios [fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip).
+
+Static model can be downloaded here [fastspeech2_nosil_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_static_0.4.zip).
 
 FastSpeech2 checkpoint contains files listed below.
 ```text
diff --git a/examples/csmsc/voc1/README.md b/examples/csmsc/voc1/README.md
index f789cba0e..b9c8a465f 100644
--- a/examples/csmsc/voc1/README.md
+++ b/examples/csmsc/voc1/README.md
@@ -13,7 +13,7 @@ Assume the path to the dataset is `~/datasets/BZNSYP`.
 Assume the path to the MFA result of CSMSC is `./baker_alignment_tone`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
@@ -123,7 +123,8 @@ optional arguments:
 
 ## Pretrained Models
 Pretrained model can be downloaded here [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip).
-Static models can be downloaded here [pwg_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_static_0.4.zip).
+
+Static model can be downloaded here [pwg_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_static_0.4.zip).
 
 Parallel WaveGAN checkpoint contains files listed below.
 
diff --git a/examples/csmsc/voc3/README.md b/examples/csmsc/voc3/README.md
index 9cb9d34d4..a72f60f16 100644
--- a/examples/csmsc/voc3/README.md
+++ b/examples/csmsc/voc3/README.md
@@ -13,7 +13,7 @@ Assume the path to the dataset is `~/datasets/BZNSYP`.
 Assume the path to the MFA result of CSMSC is `./baker_alignment_tone`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
@@ -106,8 +106,51 @@ optional arguments:
 4. `--output-dir` is the directory to save the synthesized audio files.
 5. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
 
+## Finetune
+Since there are no `noise` in the input of Multi Band MelGAN, the  audio quality is not so good (see [espnet issue](https://github.com/espnet/espnet/issues/3536#issuecomment-916035415)), we refer to the method proposed in [HiFiGAN](https://arxiv.org/abs/2010.05646),  finetune Multi Band MelGAN with the predicted mel-spectrogram from `FastSpeech2`.
+
+The length of mel-spectrograms should align with the length of wavs, so we should generate mels using ground truth alignment.
+
+But since we are fine-tuning, we should use the statistics computed during training step.
+
+You should  first download pretrained `FastSpeech2` model from [fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip) and `unzip` it.
+
+Assume the path to the dump-dir of  training  step is `dump`.
+Assume the path to the duration result of CSMSC is `durations.txt` (generated during training step's preprocessing).
+Assume the path to the pretrained `FastSpeech2` model is `fastspeech2_nosil_baker_ckpt_0.4`.
+\
+The `finetune.sh` can
+1. **source path**.
+2. generate ground truth alignment mels.
+3. link `*_wave.npy` from `dump` to `dump_finetune` (because we only use new mels, the wavs are the ones used during train step) .
+4. copy features' stats from `dump` to `dump_finetune`.
+5. normalize the ground truth alignment mels.
+6. finetune the model.
+
+Before finetune, make sure that the pretrained model is in `finetune.sh` 's `${output-dir}/checkpoints`, and there is a `records.jsonl` in it to refer to this pretrained model
+```text
+exp/finetune/checkpoints
+├── records.jsonl
+└── snapshot_iter_1000000.pdz
+```
+The content of `records.jsonl` should be as follows (change `"path"` to your own ckpt path):
+```
+{"time": "2021-11-21 15:11:20.337311", "path": "~/PaddleSpeech/examples/csmsc/voc3/exp/finetune/checkpoints/snapshot_iter_1000000.pdz", "iteration": 1000000}↩
+```
+Run the command below 
+```bash
+./finetune.sh
+```
+By default, `finetune.sh` will use `conf/finetune.yaml` as config, the dump-dir is `dump_finetune`, the experiment dir is `exp/finetune`.
+
+TODO: 
+The hyperparameter of  `finetune.yaml` is not good enough, a smaller `learning_rate` should be used (more `milestones` should be set).
+
 ## Pretrained Models
 Pretrained model can be downloaded here [mb_melgan_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_ckpt_0.5.zip).
+
+Finetuned model can ben downloaded here [mb_melgan_baker_finetune_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_finetune_ckpt_0.5.zip).
+
 Static model can be downloaded here [mb_melgan_baker_static_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_static_0.5.zip)
 
 Multi Band MelGAN checkpoint contains files listed below.
diff --git a/examples/csmsc/voc3/conf/finetune.yaml b/examples/csmsc/voc3/conf/finetune.yaml
index e02f3e220..80ab6bed7 100644
--- a/examples/csmsc/voc3/conf/finetune.yaml
+++ b/examples/csmsc/voc3/conf/finetune.yaml
@@ -128,7 +128,7 @@ discriminator_scheduler_params:
 #                    INTERVAL SETTING                     #
 ###########################################################
 discriminator_train_start_steps: 200000 # Number of steps to start to train discriminator.
-train_max_steps: 1200000                # Number of training steps.
+train_max_steps: 2000000                # Number of training steps.
 save_interval_steps: 1000              # Interval steps to save checkpoint.
 eval_interval_steps: 1000               # Interval steps to evaluate the network.
 
@@ -136,4 +136,4 @@ eval_interval_steps: 1000               # Interval steps to evaluate the network
 #                     OTHER SETTING                       #
 ###########################################################
 num_snapshots: 10                 # max number of snapshots to keep while training
-seed: 42                          # random seed for paddle, random, and np.random
\ No newline at end of file
+seed: 42                          # random seed for paddle, random, and np.random
diff --git a/examples/csmsc/voc3/finetune.sh b/examples/csmsc/voc3/finetune.sh
index 42e5a3979..4ab10e5b3 100755
--- a/examples/csmsc/voc3/finetune.sh
+++ b/examples/csmsc/voc3/finetune.sh
@@ -9,20 +9,19 @@ stop_stage=100
 source ${MAIN_ROOT}/utils/parse_options.sh || exit 1
 
 if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
-  python3 ${MAIN_ROOT}/paddlespeech/t2s/exps/fastspeech2/gen_gta_mel.py \
-      --fastspeech2-config=fastspeech2_nosil_baker_ckpt_0.4/default.yaml \
-      --fastspeech2-checkpoint=fastspeech2_nosil_baker_ckpt_0.4/snapshot_iter_76000.pdz \
-      --fastspeech2-stat=fastspeech2_nosil_baker_ckpt_0.4/speech_stats.npy \
-      --dur-file=durations.txt \
-      --output-dir=dump_finetune \
-      --phones-dict=fastspeech2_nosil_baker_ckpt_0.4/phone_id_map.txt
+    python3 ${MAIN_ROOT}/paddlespeech/t2s/exps/fastspeech2/gen_gta_mel.py \
+        --fastspeech2-config=fastspeech2_nosil_baker_ckpt_0.4/default.yaml \
+        --fastspeech2-checkpoint=fastspeech2_nosil_baker_ckpt_0.4/snapshot_iter_76000.pdz \
+        --fastspeech2-stat=fastspeech2_nosil_baker_ckpt_0.4/speech_stats.npy \
+        --dur-file=durations.txt \
+        --output-dir=dump_finetune \
+        --phones-dict=fastspeech2_nosil_baker_ckpt_0.4/phone_id_map.txt
 fi
 
 if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
-  python3 local/link_wav.py \
-    --old-dump-dir=dump \
-    --dump-dir=dump_finetune 
-
+    python3 local/link_wav.py \
+        --old-dump-dir=dump \
+        --dump-dir=dump_finetune
 fi
 
 if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
@@ -51,13 +50,13 @@ if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
 fi
 
 if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
-  CUDA_VISIBLE_DEVICES=${gpus} \
-  FLAGS_cudnn_exhaustive_search=true \
-  FLAGS_conv_workspace_size_limit=4000 \
-  python ${BIN_DIR}/train.py \
-      --train-metadata=dump_finetune/train/norm/metadata.jsonl \
-      --dev-metadata=dump_finetune/dev/norm/metadata.jsonl \
-      --config=conf/finetune.yaml \
-      --output-dir=exp/finetune \
-      --ngpu=1
+    CUDA_VISIBLE_DEVICES=${gpus} \
+    FLAGS_cudnn_exhaustive_search=true \
+    FLAGS_conv_workspace_size_limit=4000 \
+    python ${BIN_DIR}/train.py \
+        --train-metadata=dump_finetune/train/norm/metadata.jsonl \
+        --dev-metadata=dump_finetune/dev/norm/metadata.jsonl \
+        --config=conf/finetune.yaml \
+        --output-dir=exp/finetune \
+        --ngpu=1
 fi 
\ No newline at end of file
diff --git a/examples/ljspeech/tts0/README.md b/examples/ljspeech/tts0/README.md
index a3e947e81..09fd0c133 100644
--- a/examples/ljspeech/tts0/README.md
+++ b/examples/ljspeech/tts0/README.md
@@ -12,7 +12,7 @@ tar xjvf LJSpeech-1.1.tar.bz2
 Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize mels.
 ```bash
diff --git a/examples/ljspeech/tts1/README.md b/examples/ljspeech/tts1/README.md
index fbb2c9d04..12e43e2ea 100644
--- a/examples/ljspeech/tts1/README.md
+++ b/examples/ljspeech/tts1/README.md
@@ -12,7 +12,7 @@ tar xjvf LJSpeech-1.1.tar.bz2
 Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
diff --git a/examples/ljspeech/tts3/README.md b/examples/ljspeech/tts3/README.md
index bc38aac64..cda535412 100644
--- a/examples/ljspeech/tts3/README.md
+++ b/examples/ljspeech/tts3/README.md
@@ -14,7 +14,7 @@ Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
 Assume the path to the MFA result of LJSpeech-1.1 is `./ljspeech_alignment`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
diff --git a/examples/ljspeech/voc0/README.md b/examples/ljspeech/voc0/README.md
index ad2337ef9..09856c367 100644
--- a/examples/ljspeech/voc0/README.md
+++ b/examples/ljspeech/voc0/README.md
@@ -13,7 +13,7 @@ Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
 Assume the path to the Tacotron2 generated mels is `../tts0/output/test`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs from mels.
 ```bash
diff --git a/examples/ljspeech/voc1/README.md b/examples/ljspeech/voc1/README.md
index fdeac6329..0506d5d8f 100644
--- a/examples/ljspeech/voc1/README.md
+++ b/examples/ljspeech/voc1/README.md
@@ -12,7 +12,7 @@ Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
 Assume the path to the MFA result of LJSpeech-1.1 is `./ljspeech_alignment`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
diff --git a/examples/vctk/tts3/README.md b/examples/vctk/tts3/README.md
index 78bfb9668..334372f95 100644
--- a/examples/vctk/tts3/README.md
+++ b/examples/vctk/tts3/README.md
@@ -17,7 +17,7 @@ Assume the path to the dataset is `~/datasets/VCTK-Corpus-0.92`.
 Assume the path to the MFA result of VCTK is `./vctk_alignment`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.
diff --git a/examples/vctk/voc1/README.md b/examples/vctk/voc1/README.md
index c9ecae0d2..5063b869c 100644
--- a/examples/vctk/voc1/README.md
+++ b/examples/vctk/voc1/README.md
@@ -17,7 +17,7 @@ Assume the path to the dataset is `~/datasets/VCTK-Corpus-0.92`.
 Assume the path to the MFA result of VCTK is `./vctk_alignment`.
 Run the command below to
 1. **source path**.
-2. preprocess the dataset,
+2. preprocess the dataset.
 3. train the model.
 4. synthesize wavs.
     - synthesize waveform from `metadata.jsonl`.