2021-08-28 19:56:19 +08:00
2021-08-07 11:56:00 +08:00
[](http://choosealicense.com/licenses/mit/)
2021-08-23 15:37:33 +08:00
> English | [中文](README-CN.md)
2021-08-07 11:56:00 +08:00
## Features
2021-08-28 19:56:19 +08:00
🌍 **Chinese** supported mandarin and tested with multiple datasets: aidatatang_200zh, magicdata, aishell3
2021-08-07 11:56:00 +08:00
🤩 **PyTorch** worked for pytorch, tested in version of 1.9.0(latest in August 2021), with GPU Tesla T4 and GTX 2060
2021-08-23 15:37:33 +08:00
🌍 **Windows + Linux** tested in both Windows OS and linux OS after fixing nits
2021-08-07 11:56:00 +08:00
🤩 **Easy & Awesome** effect with only newly-trained synthesizer, by reusing the pretrained encoder/vocoder
2021-08-07 16:36:07 +08:00
### [DEMO VIDEO](https://www.bilibili.com/video/BV1sA411P7wM/)
2021-08-07 11:56:00 +08:00
## Quick Start
### 1. Install Requirements
> Follow the original repo to test if you got all environment ready.
**Python 3.7 or higher ** is needed to run the toolbox.
* Install [PyTorch ](https://pytorch.org/get-started/locally/ ).
2021-08-23 09:17:32 +08:00
> If you get an `ERROR: Could not find a version that satisfies the requirement torch==1.9.0+cu102 (from versions: 0.1.2, 0.1.2.post1, 0.1.2.post2 )` This error is probably due to a low version of python, try using 3.9 and it will install successfully
2021-08-07 11:56:00 +08:00
* Install [ffmpeg ](https://ffmpeg.org/download.html#get-packages ).
* Run `pip install -r requirements.txt` to install the remaining necessary packages.
2021-08-23 15:37:33 +08:00
* Install webrtcvad `pip install webrtcvad-wheels` (If you need)
2021-08-16 22:18:46 +08:00
> Note that we are using the pretrained encoder/vocoder but synthesizer, since the original model is incompatible with the Chinese sympols. It means the demo_cli is not working at this moment.
### 2. Train synthesizer with your dataset
2021-08-25 23:14:30 +08:00
* Download aidatatang_200zh or other dataset and unzip: make sure you can access all .wav in *train* folder
2021-08-07 11:56:00 +08:00
* Preprocess with the audios and the mel spectrograms:
`python synthesizer_preprocess_audio.py <datasets_root>`
2021-08-25 23:14:30 +08:00
Allow parameter `--dataset {dataset}` to support adatatang_200zh, magicdata, aishell3
2021-08-23 15:37:33 +08:00
>If it happens `the page file is too small to complete the operation` , please refer to this [video ](https://www.youtube.com/watch?v=Oh6dga-Oy10&ab_channel=CodeProf ) and change the virtual memory to 100G (102400), for example : When the file is placed in the D disk, the virtual memory of the D disk is changed.
2021-08-07 11:56:00 +08:00
* Preprocess the embeddings:
`python synthesizer_preprocess_embeds.py <datasets_root>/SV2TTS/synthesizer`
* Train the synthesizer:
`python synthesizer_train.py mandarin <datasets_root>/SV2TTS/synthesizer`
2021-08-23 15:37:33 +08:00
* Go to next step when you see attention line show and loss meet your need in training folder *synthesizer/saved_models/* .
2021-08-07 11:56:00 +08:00
> FYI, my attention came after 18k steps and loss became lower than 0.4 after 50k steps.
2021-08-07 12:02:34 +08:00
2021-08-22 20:37:34 +08:00
### 2.2 Use pretrained model of synthesizer
> Thanks to the community, some models will be shared:
2021-08-23 15:37:33 +08:00
| author | Download link | Previow Video |
| --- | ----------- | ----- |
2021-08-22 20:37:34 +08:00
|@miven| https://pan.baidu.com/s/1PI-hM3sn5wbeChRryX-RCQ code：
2021-08-16 09:35:09 +08:00
> A link to my early trained model: [Baidu Yun](https://pan.baidu.com/s/10t3XycWiNIg5dN5E_bMORQ)
Code：
2021-08-29 15:48:11 +08:00
### 2.3 Train vocoder (Optional)
* Preprocess the data:
`python vocoder_preprocess.py <datasets_root>`
* Train the vocoder:
`python vocoder_train.py mandarin <datasets_root>`
2021-08-16 22:18:46 +08:00
### 3. Launch the Toolbox
2021-08-07 11:56:00 +08:00
You can then try the toolbox:
2021-08-23 15:37:33 +08:00
`python demo_toolbox.py -d <datasets_root>`
or
`python demo_toolbox.py`
2021-08-07 11:56:00 +08:00
2021-08-16 22:22:55 +08:00
> Good news🤩: Chinese Characters are supported
2021-08-07 11:56:00 +08:00
## TODO
2021-08-07 16:36:07 +08:00
- [x] Add demo video
2021-08-13 00:41:58 +08:00
- [X] Add support for more dataset
2021-08-16 09:35:09 +08:00
- [X] Upload pretrained model
2021-08-16 22:18:46 +08:00
- [ ] Support parallel tacotron
- [ ] Service orianted and docterize
2021-08-07 11:56:00 +08:00
- 🙏 Welcome to add more
2021-08-29 10:55:59 +08:00
## Reference
> This repository is forked from [Real-Time-Voice-Cloning](https://github.com/CorentinJ/Real-Time-Voice-Cloning) which only support English.
| URL | Designation | Title | Implementation source |
| --- | ----------- | ----- | --------------------- |
|[**1806.04558**](https://arxiv.org/pdf/1806.04558.pdf) | **SV2TTS** | **Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis** | This repo |
|[1802.08435](https://arxiv.org/pdf/1802.08435.pdf) | WaveRNN (vocoder) | Efficient Neural Audio Synthesis | [fatchord/WaveRNN ](https://github.com/fatchord/WaveRNN ) |
|[1703.10135](https://arxiv.org/pdf/1703.10135.pdf) | Tacotron (synthesizer) | Tacotron: Towards End-to-End Speech Synthesis | [fatchord/WaveRNN ](https://github.com/fatchord/WaveRNN )
|[1710.10467](https://arxiv.org/pdf/1710.10467.pdf) | GE2E (encoder)| Generalized End-To-End Loss for Speaker Verification | This repo |
