Skip to content

Commit

Permalink
add binarized dataset link
Browse files Browse the repository at this point in the history
  • Loading branch information
yerfor committed May 12, 2022
1 parent d528377 commit 6f5fed7
Show file tree
Hide file tree
Showing 3 changed files with 7 additions and 27 deletions.
4 changes: 2 additions & 2 deletions README-zh.md
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@

1. 我们提出了**Syntactic Graph Builder (论文的3.1小节)****Syntactic Graph Encoder (论文的3.2小节)**,被证明是提取句法特征以提高韵律建模和持续时间准确性的有效单元 TTS 模型。
2. 我们引入了**Multi-Length Adversarial Training (论文的3.3小节)**,它可以替代PortaSpeech 中基于flow的post-net,加快推理时间的同时提高音频质量的自然度。
3. 我们支持三个数据集:[LJSpeech](https://keithito.com/LJ-Speech-Dataset/)(单人英语数据集)、[Biaobei](https://www.data-baker.com/ open%20source.html)(单人中文数据集)和[LibriTTS](http://www.openslr.org/60)(多人英语数据集)
3. 我们支持三个数据集:[LJSpeech](https://keithito.com/LJ-Speech-Dataset/)(单人英语数据集)、[Biaobei]() (单人中文数据集)和[LibriTTS](http://www.openslr.org/60)(多人英语数据集)

搭建环境

Expand All @@ -40,7 +40,7 @@ bash mfa_usr/install_mfa.sh # install force alignment tools

#### 准备数据集

您可以直接使用我们的处理好的LJSpeech数据集和 Biaobei数据集。 从[这个链接]() 下载它们并将它们解压缩到 `data/binary/` 文件夹中。
您可以直接使用我们的处理好的[LJSpeech数据集](https://drive.google.com/file/d/1WfErAxKqMluQU3vupWS6VB6NdehXwCKM/view?usp=sharing)[Biaobei](https://drive.google.com/file/d/1-ApEbBrW5kfF0jM18EmW7DCsll-c1ROp/view?usp=sharing)数据集。 从链接给的谷歌云盘里下载它们并将它们解压缩到 `data/binary/` 文件夹中。

至于 LibriTTS,您可以下载原始数据集并使用我们的“data_gen”模块对其进行处理。 详细说明可以在 [dosc/prepare_data](docs/prepare_data.md) 中找到。

Expand Down
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,7 @@ bash mfa_usr/install_mfa.sh # install force alignment tools

#### Data Preparation

You can directly use our binarized datasets for LJSpeech and Biaobei. Download them from [this link]() and unzip them into the `data/binary/` folder.
You can directly use our binarized datasets for [LJSpeech](https://drive.google.com/file/d/1WfErAxKqMluQU3vupWS6VB6NdehXwCKM/view?usp=sharing) and [Biaobei](https://drive.google.com/file/d/1-ApEbBrW5kfF0jM18EmW7DCsll-c1ROp/view?usp=sharing). Download them and unzip them into the `data/binary/` folder.

As for LibriTTS, you can download the raw datasets and process them with our `data_gen` modules. Detailed instructions can be found in [dosc/prepare_data](docs/prepare_data.md).

Expand Down
28 changes: 4 additions & 24 deletions docs/prepare_data.md
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,10 @@ python data_gen/tts/runs/train_mfa_align.py --config egs/lj/base_text2mel.yaml
python data_gen/tts/runs/binarize.py --config egs/lj/base_text2mel.yaml
```

## Biaobei

Please download the binarized dataset from [this link]()

## LibriTTS

### Download Dataset
Expand All @@ -47,28 +51,4 @@ python data_gen/tts/runs/train_mfa_align.py --config egs/libritts/base_text2mel.
python data_gen/tts/runs/binarize.py --config egs/libritts/base_text2mel.yaml
```

## LJSpeech

### Download Dataset

```bash
mkdir -p data/raw/ljspeech
cd data/raw
wget https://data.keithito.com/data/speech/LJSpeech-1.1.tar.bz2
bzip2 -d LJSpeech-1.1.tar.bz2
tar -xvf LJSpeech-1.1.tar
cd ../../
```

### Forced Align and Preprocess Dataset

```bash
# Preprocess step: text and unify the file structure.
python data_gen/tts/runs/preprocess.py --config $CONFIG_NAME
# Align step: MFA alignment.
python data_gen/tts/runs/train_mfa_align.py --config $CONFIG_NAME
# Binarization step: Binarize data for fast IO. You only need to rerun this line when running different task if you have `preprocess`ed and `align`ed the dataset before.
python data_gen/tts/runs/binarize.py --config $CONFIG_NAME
```

## More datasets will be supported soon...

0 comments on commit 6f5fed7

Please sign in to comment.