TTS ‐ supertonic

Supertonic is a TTS system that focuses on generating high-quality, natural-sounding speech. It utilizes advanced deep learning techniques to produce realistic voice output, making it suitable for various applications such as virtual assistants, audiobooks, and accessibility tools. Supertonic aims to provide a seamless and immersive audio experience for users by leveraging state-of-the-art models and algorithms in the field of text-to-speech synthesis.

More details can be found in the Supertonic GitHub repository

Install Supertonic

Before cloning the repository, make sure you have git lfs (Large File Storage) installed on your system, as the Supertonic assets include large model files that require git lfs to download properly.

sudo apt-get update
sudo apt-get install git-lfs
git lfs install

Whisplay Chatbot uses the node runtime to run Supertonic, all you need is to pull the supertonic assets repository:

cd
git clone https://huggingface.co/Supertone/supertonic-2

Configure Whisplay AI Chatbot to use Supertonic

.env file settings:

TTS_SERVER=supertonic

## Supertonic TTS
# Supertonic is an ultra-fast, lightweight, on-device TTS engine powered by ONNX Runtime
# GitHub: https://github.com/supertone-inc/supertonic
# Features: 167x faster than real-time, 66M parameters, multilingual support (en, ko, es, pt, fr)
# Download models first:
# git clone https://huggingface.co/Supertone/supertonic-2 assets
# The assets folder should contain:
# - assets/onnx/ (model files: dp.onnx, text_enc.onnx, vector_est.onnx, vocoder.onnx, config.json, unicode_lookup.json)
# - assets/voice_styles/ (voice style files: M1.json - M5.json, F1.json - F5.json)
# 
# Directory where Supertonic assets are stored (default: ./assets)
SUPERTONIC_ASSETS_DIR=/home/pi/supertonic-2
# Voice style to use: M1-M5 (male voices), F1-F5 (female voices), default is M1
# SUPERTONIC_VOICE_STYLE=M1
# Language code: en (English), ko (Korean), es (Spanish), pt (Portuguese), fr (French), default is en
# SUPERTONIC_LANGUAGE=en
# Number of denoising steps (higher = better quality but slower), default is 5
# SUPERTONIC_TOTAL_STEP=5
# Speech speed multiplier (higher = faster speech), default is 1.05
# SUPERTONIC_SPEED=1.05
# Silence duration between text chunks in seconds, default is 0.3
# SUPERTONIC_SILENCE_DURATION=0.3

Restart the chatbot after updating the .env file, and it will use Supertonic for TTS synthesis. You can adjust the voice style, language, speed, and other parameters by modifying the corresponding environment variables in the .env file.