sovaai / sova-tts

SOVA TTS

SOVA TTS is a speech syntthesis solution based on Tacotron 2 architecture. It is designed as a REST API service and it can be customized (both code and models) for your needs.

Installation

The easiest way to deploy the service is via docker-compose, so you have to install Docker and docker-compose first. Here's a brief instruction for Ubuntu:

Docker installation

• Install Docker:

$ sudo apt-get update
$ sudo apt-get install \
    apt-transport-https \
    ca-certificates \
    curl \
    gnupg-agent \
    software-properties-common
$ curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
$ sudo apt-key fingerprint 0EBFCD88
$ sudo add-apt-repository \
   "deb [arch=amd64] https://download.docker.com/linux/ubuntu \
   $(lsb_release -cs) \
   stable"
$ sudo apt-get update
$ sudo apt-get install docker-ce docker-ce-cli containerd.io
$ sudo usermod -aG docker $(whoami)

In order to run docker commands without sudo you might need to relogin.

• Install docker-compose:

$ sudo curl -L "https://github.com/docker/compose/releases/download/1.25.5/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
$ sudo chmod +x /usr/local/bin/docker-compose

• (Optional) If you're planning on using CUDA run these commands:

$ curl -s -L https://nvidia.github.io/nvidia-container-runtime/gpgkey | \
  sudo apt-key add -
$ distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
$ curl -s -L https://nvidia.github.io/nvidia-container-runtime/$distribution/nvidia-container-runtime.list | \
  sudo tee /etc/apt/sources.list.d/nvidia-container-runtime.list
$ sudo apt-get update
$ sudo apt-get install nvidia-container-runtime

Add the following content to the file /etc/docker/daemon.json:

{
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    },
    "default-runtime": "nvidia"
}

Restart the service:

$ sudo systemctl restart docker.service

Build and deploy

• Clone the repository, download the pretrained models archive and extract the contents into the project folder:

$ git clone --recursive https://github.com/sovaai/sova-tts.git --branch v1.1
$ cd sova-tts/
$ wget http://dataset.sova.ai/SOVA-TTS/Data_v1.1.tar
$ tar -xvf Data_v1.1.tar && rm Data_v1.1.tar

• Build docker image

  • Build sova-tts-gpu image if you're planning on using GPU:

  $ sudo docker-compose build sova-tts-gpu

  • Build sova-tts image if you're planning on using CPU:

  $ sudo docker-compose build sova-tts

• Run the desired service container

  • GPU:

  $ sudo docker-compose up -d sova-tts-gpu

  • CPU:

  $ sudo docker-compose up -d sova-tts

Testing

To test the service you can send a POST request:

$ curl --request POST 'http://localhost:8899/synthesize/' \
--header 'Content-Type: application/json' \
--data-raw '{
    "text": "Добрый день! Как ваши дел+а?",
    "voice": "Natasha"
}'

Acknowledgements

Original Tacotron 2 implementation by NVIDIA.