- VITS text-to-speech, voice conversion
- HuBert-soft VITS
- vits_chinese
- Bert-VITS2
- W2V2 VITS / emotional-vits dimensional emotion model
- Support for loading multiple models
- Automatic language recognition and processing,set the scope of language type recognition according to model's cleaner,support for custom language type range
- Customize default parameters
- Long text batch processing
- GPU accelerated inference
- SSML (Speech Synthesis Markup Language) work in progress...
Please note that different IDs may support different languages.speakers
https://artrajz-vits-simple-api.hf.space/voice/vits?text=你好,こんにちは&id=164
https://artrajz-vits-simple-api.hf.space/voice/vits?text=Difficult the first time, easy the second.&id=4
- excited:
https://artrajz-vits-simple-api.hf.space/voice/w2v2-vits?text=こんにちは&id=3&emotion=111
- whispered:
https://artrajz-vits-simple-api.hf.space/w2v2-vits?text=こんにちは&id=3&emotion=2077
ssml.mov
There are two deployment options to choose from. Regardless of the option you select, you'll need to import the model after deployment to use the application.
Run the following command to pull the Docker image. Follow the prompts in the script to choose the necessary files to download and pull the image:
bash -c "$(wget -O- https://raw.githubusercontent.com/Artrajz/vits-simple-api/main/vits-simple-api-installer-latest.sh)"
The default paths for project configuration files and model folders are /usr/local/vits-simple-api/
.
Run the following command to start the container:
docker-compose up -d
To update the image, run the following commands:
docker-compose pull
Then, restart the container:
docker-compose up -d
Clone the project repository using the following command:
git clone https://github.com/Artrajz/vits-simple-api.git
It's recommended to use a Python virtual environment. Run the following command to install the required Python dependencies:
pip install -r requirements.txt
Run the following command to start the program:
python app.py
Go to the releases page and download the latest deployment package. Extract the downloaded files.
Run start.bat
to launch the program.
Download the VITS model files and place them in the Model directory.
If you are starting for the first time, modify the default model path configuration in the config.py file (optional).
After the first startup, a config.yml configuration file will be generated. You can either modify the model_list in the configuration file or make changes through the admin backend in the browser.
You can specify the model paths using either absolute or relative paths, where relative paths are considered from the Model folder in the project's root directory.
For example, if the Model folder contains the following files:
├─model1
│ │─G_1000.pth
│ └─config.json
└─model2
│─G_1000.pth
└─config.json
You have multiple options for specifying the paths based on your preference.
Option 1:
'model_config':
'model_list':
- - model1/G_1000.pth
- model1/config.json
- - model2/G_1000.pth
- model2/config.json
Option 2:
'model_config':
'model_list':
- [model1/G_1000.pth, model1/config.json]
- [model2/G_1000.pth, model2/config.json]
Option 3:
'model_config':
'model_list': [
[model1/G_1000.pth, model1/config.json],
[model2/G_1000.pth, model2/config.json],
]
Check the highest version of CUDA supported by your graphics card:
nvidia-smi
Taking CUDA 11.7 as an example, download it from the official website
1.13.1+cu117 is recommended, other versions may have memory instability issues.
pip install torch==1.13.1+cu117 --extra-index-url https://download.pytorch.org/whl/cu117
The installation process is similar, but I don't have the environment to test it.
*Port is modifiable under the default setting of port 23456.
The default address is http://127.0.0.1:23456/admin.
The initial username and password can be found at the bottom of the config.yml file after the first startup.
The admin backend allows loading and unloading models, and while it has login authentication, for added security, you can disable the admin backend in the config.yml
:
'IS_ADMIN_ENABLED': !!bool 'false'
This extra measure helps ensure absolute security when making the admin backend inaccessible to the public network.
Starting from Bert-VITS2 v2.0, a model requires loading three different language Bert models. If you only need to use one or two languages, you can add the lang
parameter in the config.json
file of the model's data section. The value ['zh']
indicates that the model only uses Chinese and will load Chinese Bert models. The value ['zh', 'ja']
indicates the usage of both Chinese and Japanese bilingual models, and only Chinese and Japanese Bert models will be loaded. Similarly, this pattern continues for other language combinations.
Example:
"data": {
"lang": ["zh", "ja"],
"training_files": "filelists/train.list",
"validation_files": "filelists/val.list",
"max_wav_value": 32768.0,
...
Fasttext may not be installed on windows, you can install it with the following command,or download wheels here
# For Python 3.10 on win_amd64
pip install https://github.com/Artrajz/archived/raw/main/fasttext/fasttext-0.9.2-cp310-cp310-win_amd64.whl
or
pip install fasttext -i https://pypi.artrajz.cn/simple
Since pypi.org does not provide a wheel file for pyopenjtalk, you often need to install it from the source code. This process might be cumbersome for some users, so you can also install it using a pre-built wheel as follows:
pip install pyopenjtalk -i https://pypi.artrajz.cn/simple
To ensure compatibility with the Bert-VITS2 model, modify the config.json file by adding a version parameter "version": "x.x.x". For instance, if the model version is 1.0.1, the configuration file should be written as:
{
"version": "1.0.1",
"train": {
"log_interval": 10,
"eval_interval": 100,
"seed": 52,
...
-
GET http://127.0.0.1:23456/voice/speakers
Returns the mapping table of role IDs to speaker names.
-
GET http://127.0.0.1:23456/voice/vits?text=text
Default values are used when other parameters are not specified.
-
GET http://127.0.0.1:23456/voice/vits?text=[ZH]text[ZH][JA]text[JA]&lang=mix
When lang=mix, the text needs to be annotated.
-
GET http://127.0.0.1:23456/voice/vits?text=text&id=142&format=wav&lang=zh&length=1.4
The text is "text", the role ID is 142, the audio format is wav, the text language is zh, the speech length is 1.4, and the other parameters are default.
- See
api_test.py
Set API_KEY_ENABLED = True
in config.py
to enable API key authentication. The API key is API_KEY = "api-key"
.
After enabling it, you need to add the api_key
parameter in GET requests and add the X-API-KEY
parameter in the header for POST requests.
Name | Parameter | Is must | Default | Type | Instruction |
---|---|---|---|---|---|
Synthesized text | text | true | str | Text needed for voice synthesis. | |
Speaker ID | id | false | From config.yml |
int | The speaker ID. |
Audio format | format | false | From config.yml |
str | Support for wav,ogg,silk,mp3,flac |
Text language | lang | false | From config.yml |
str | The language of the text to be synthesized. Available options include auto, zh, ja, and mix. When lang=mix, the text should be wrapped in [ZH] or [JA].The default mode is auto, which automatically detects the language of the text |
Audio length | length | false | From config.yml |
float | Adjusts the length of the synthesized speech, which is equivalent to adjusting the speed of the speech. The larger the value, the slower the speed. |
Noise | noise | false | From config.yml |
float | Sample noise, controlling the randomness of the synthesis. |
SDP noise | noisew | false | From config.yml |
float | Stochastic Duration Predictor noise, controlling the length of phoneme pronunciation. |
Segment Size | segment_size | false | From config.yml |
int | Divide the text into paragraphs based on punctuation marks, and combine them into one paragraph when the length exceeds segment_size. If segment_size<=0, the text will not be divided into paragraphs. |
Streaming response | streaming | false | false | bool | Streamed synthesized speech with faster initial response. |
Name | Parameter | Is must | Default | Type | Instruction |
---|---|---|---|---|---|
Uploaded Audio | upload | true | file | The audio file to be uploaded. It should be in wav or ogg | |
Source Role ID | original_id | true | int | The ID of the role used to upload the audio file. | |
Target Role ID | target_id | true | int | The ID of the target role to convert the audio to. |
Name | Parameter | Is must | Default | Type | Instruction |
---|---|---|---|---|---|
Uploaded Audio | upload | true | file | The audio file to be uploaded. It should be in wav or ogg format. | |
Target speaker ID | id | true | int | The target speaker ID. | |
Audio format | format | true | str | wav,ogg,silk | |
Audio length | length | true | float | Adjusts the length of the synthesized speech, which is equivalent to adjusting the speed of the speech. The larger the value, the slower the speed. | |
Noise | noise | true | float | Sample noise, controlling the randomness of the synthesis. | |
sdp noise | noisew | true | float | Stochastic Duration Predictor noise, controlling the length of phoneme pronunciation. |
Name | Parameter | Is must | Default | Type | Instruction |
---|---|---|---|---|---|
Synthesized text | text | true | str | Text needed for voice synthesis. | |
Speaker ID | id | false | From config.yml |
int | The speaker ID. |
Audio format | format | false | From config.yml |
str | Support for wav,ogg,silk,mp3,flac |
Text language | lang | false | From config.yml |
str | The language of the text to be synthesized. Available options include auto, zh, ja, and mix. When lang=mix, the text should be wrapped in [ZH] or [JA].The default mode is auto, which automatically detects the language of the text |
Audio length | length | false | From config.yml |
float | Adjusts the length of the synthesized speech, which is equivalent to adjusting the speed of the speech. The larger the value, the slower the speed. |
Noise | noise | false | From config.yml |
float | Sample noise, controlling the randomness of the synthesis. |
SDP noise | noisew | false | From config.yml |
float | Stochastic Duration Predictor noise, controlling the length of phoneme pronunciation. |
Segment Size | segment_size | false | From config.yml |
int | Divide the text into paragraphs based on punctuation marks, and combine them into one paragraph when the length exceeds segment_size. If segment_size<=0, the text will not be divided into paragraphs. |
Dimensional emotion | emotion | false | 0 | int | The range depends on the emotion reference file in npy format, such as the range of the innnky's model all_emotions.npy, which is 0-5457. |
Name | Parameter | Is must | Default | Type | Instruction |
---|---|---|---|---|---|
Uploaded Audio | upload | true | file | Return the npy file that stores the dimensional emotion vectors. |
Name | Parameter | Is must | Default | Type | Instruction |
---|---|---|---|---|---|
Synthesized text | text | true | str | Text needed for voice synthesis. | |
Speaker ID | id | false | From config.yml |
int | The speaker ID. |
Audio format | format | false | From config.yml |
str | Support for wav,ogg,silk,mp3,flac |
Text language | lang | false | From config.yml |
str | "Auto" is a mode for automatic language detection and is also the default mode. However, it currently only supports detecting the language of an entire text passage and cannot distinguish languages on a per-sentence basis. The other available language options are "zh" and "ja". |
Audio length | length | false | From config.yml |
float | Adjusts the length of the synthesized speech, which is equivalent to adjusting the speed of the speech. The larger the value, the slower the speed. |
Noise | noise | false | From config.yml |
float | Sample noise, controlling the randomness of the synthesis. |
SDP noise | noisew | false | From config.yml |
float | Stochastic Duration Predictor noise, controlling the length of phoneme pronunciation. |
Segment Size | segment_size | false | From config.yml |
int | Divide the text into paragraphs based on punctuation marks, and combine them into one paragraph when the length exceeds segment_size. If segment_size<=0, the text will not be divided into paragraphs. |
SDP/DP mix ratio | sdp_ratio | false | From config.yml |
int | The theoretical proportion of SDP during synthesis, the higher the ratio, the larger the variance in synthesized voice tone. |
Emotion | emotion | false | None | int | Available for Bert-VITS2 v2.1, ranging from 0 to 9 |
Emotion reference Audio | reference_audio | false | None | Bert-VITS2 v2.1 uses reference audio to control the synthesized audio's emotion | |
Text Prompt | text_prompt | false | None | str | Bert-VITS2 v2.2 text prompt used for emotion control |
Style Text | style_text | false | None | str | Bert-VITS2 v2.3 text prompt used for emotion control |
Style Text Weight | style_weight | false | From config.yml |
float | Bert-VITS2 v2.3 text prompt weight used for prompt weighting |
Supported Elements and Attributes
speak
Element
Attribute | Instruction | Is must |
---|---|---|
id | Default value is retrieved From config.yml |
false |
lang | Default value is retrieved From config.yml |
false |
length | Default value is retrieved From config.yml |
false |
noise | Default value is retrieved From config.yml |
false |
noisew | Default value is retrieved From config.yml |
false |
segment_size | Splits text into segments based on punctuation marks. When the sum of segment lengths exceeds segment_size , it is treated as one segment. segment_size<=0 means no segmentation. The default value is 0. |
false |
model_type | Default is VITS. Options: W2V2-VITS, BERT-VITS2 | false |
emotion | Only effective when using W2V2-VITS . The range depends on the npy emotion reference file. | false |
sdp_ratio | Only effective when using BERT-VITS2 . | false |
voice
Element
Higher priority than speak
.
Attribute | Instruction | Is must |
---|---|---|
id | Default value is retrieved From config.yml |
false |
lang | Default value is retrieved From config.yml |
false |
length | Default value is retrieved From config.yml |
false |
noise | Default value is retrieved From config.yml |
false |
noisew | Default value is retrieved From config.yml |
false |
segment_size | Splits text into segments based on punctuation marks. When the sum of segment lengths exceeds segment_size , it is treated as one segment. segment_size<=0 means no segmentation. The default value is 0. |
false |
model_type | Default is VITS. Options: W2V2-VITS, BERT-VITS2 | false |
emotion | Only effective when using W2V2-VITS . The range depends on the npy emotion reference file. | false |
sdp_ratio | Only effective when using BERT-VITS2 . | false |
break
Element
Attribute | Instruction | Is must |
---|---|---|
strength | x-weak, weak, medium (default), strong, x-strong | false |
time | The absolute duration of a pause in seconds (such as 2s ) or milliseconds (such as 500ms ). Valid values range from 0 to 5000 milliseconds. If you set a value greater than the supported maximum, the service will use 5000ms . If the time attribute is set, the strength attribute is ignored. |
false |
Strength | Relative Duration |
---|---|
x-weak | 250 ms |
weak | 500 ms |
medium | 750 ms |
strong | 1000 ms |
x-strong | 1250 ms |
Example
See api_test.py
Learning and communication,now there is only Chinese QQ group
- vits:https://github.com/jaywalnut310/vits
- MoeGoe:https://github.com/CjangCjengh/MoeGoe
- emotional-vits:https://github.com/innnky/emotional-vits
- vits-uma-genshin-honkai:https://huggingface.co/spaces/zomehwh/vits-uma-genshin-honkai
- vits_chinese:https://github.com/PlayVoice/vits_chinese
- Bert_VITS2:https://github.com/fishaudio/Bert-VITS2