piper_tts

An extension for the text-generation-webui by oobabooga that uses Piper for fast voice generation.

This project is a Web user interface (WebUI) for text generation using Gradio and a Piper text-to-speech (TTS) model. The main objective is to provide a user-friendly experience for text generation with audio.

Features

16/11/2023 -- Speaker ID :** Some model may contain several voices, so to find out which ID to use, refer to the model's JSON file.
16/11/2023 -- Sentence silence :** allows you to specify the duration, in seconds, of silence to be added after each sentence during text-to-speech.
Enable/Disable :** Enable or disable the TTS extension.
Autoplay :** Choose to automatically read generated text.
Text display :** Choose to show or hide generated text.
Custom settings :** Adjust audio parameters such as noise, phoneme length and noise width.
Template selection :** Choose from different templates available for text generation.
WAV save :** Audio files are saved in the outputs folder.
Save settings :** Save your settings.
Remove WAV :** delete all WAV files from the directory to free up storage space.

Saved settings

Selected settings are saved in a JSON file settings.json so that the user can retrieve his preferences each time he uses the device.

Initial configuration

Make sure you install all necessary dependencies and configure your environment according to the project instructions.

Installation

Clone the repository in the extensions directory.

git clone https://github.com/tijo95/piper_tts.git

download the appropriate binary for your platform from piper repository:

For windows, download https://github.com/rhasspy/piper/releases/download/2023.11.14-2/piper_windows_amd64.zip Unzip all contents into piper_tts

For linux:

cd piper_tts/
wget https://github.com/rhasspy/piper/releases/download/2023.11.14-2/piper_linux_x86_64.tar.gz
tar -xvf piper_linux_x86_64.tar.gz
rm piper_linux_x86_64.tar.gz

Download the .onnx model and their .json files and place them in the piper_tts/model directory.

The models are available at this address: https://huggingface.co/rhasspy/piper-voices/tree/v1.0.0
Run the main script and have fun surprising your AI.

Contributions

Contributions are welcome! Feel free to open an issue or propose an extraction request to improve this project.

Piper Github

Github : https://github.com/rhasspy/piper#running-in-python

Listen to voice samples : https://rhasspy.github.io/piper-samples

tijo95 / piper_tts