https://github.com/kfatehi/persian-tts-server

Simple HTTP server wrapping Coqui TTS
https://github.com/kfatehi/persian-tts-server

Last synced: 2 days ago
JSON representation

Simple HTTP server wrapping Coqui TTS

• Host: GitHub
• URL: https://github.com/kfatehi/persian-tts-server
• Owner: kfatehi
• Created: 2023-05-07T06:21:10.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2023-05-08T00:54:01.000Z (over 1 year ago)
• Last Synced: 2024-10-28T16:21:29.197Z (15 days ago)
• Language: Python
• Size: 5.86 KB
• Stars: 2
• Watchers: 2
• Forks: 1
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# persian-tts-server

Every so often I look for Persian TTS software.

This year was a huge success.

First I found https://github.com/shenasa-ai/persian-tts but I did not see any pretrained models.

Then I (re-found) https://github.com/karim23657/Persian-tts-coqui which has pretrained models!

Indeed Karim's models sound quite good! Specifically I enjoy using `persian-tts-female-vits` and this repo contains the scripts to use it.

This server is designed to be used by https://github.com/kfatehi/riva_tts_proxy/tree/sapi5_and_coqui which is designed to be used from the [Read Aloud browser extension](https://github.com/ken107/read-aloud) thus providing a high-quality, offline, easy-to-use Persian TTS solution!

## Prequisites

* Docker

* Nvidia GPU

Coqui TTS does not require using the GPU but I have made the assumption that we'll be using Nvidia GPUs through Docker.

## Get the model

```bash

mkdir -p pretrained_models

pushd pretrained_models

GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/Kamtera/persian-tts-female-vits

pushd persian-tts-female-vits

git lfs pull

popd

popd

```

## Build and run the server

```bash

docker build -t persian-tts .

docker run --rm --gpus all -v $PWD/pretrained_models/persian-tts-female-vits:/kaggle/working -p 5000:5000 -it persian-tts

```

## Customize and run forever

Here is how I use it:

- daemonized

- auto-restart on bootup unless i stop the container

- i use port 5001 because 5000 is already dedicated to my Nvidia Riva TTS proxy

```bash

docker run -d --restart=unless-stopped --gpus all -v $PWD/pretrained_models/persian-tts-female-vits:/kaggle/working -p 5001:5000 -it persian-tts

```