bruceunx/TTS-Clone-Inference

TTS with voice clone

TTS-Clone-Inference: Clean Inference and Performance Improvements

This repository contains a modified version of a TTS from coqui-ai that focuses on optimizing the inference pipeline for clean output and improved performance.

- parse inference pipeline without training or other parts

Installation

- use poetry to install dependencies

Download models

• download model from https://huggingface.co/bruceunx/tts-clone-inference/tree/main

change code in main.py

tts = TTS(model_path="models", config_path="models/config.json", gpu=False)

tts.tts_to_file(text="hello world",
                file_path="./tmp/sample.wav",
                speaker_wav="./tmp/output.wav",
                enable_text_splitting=True,
                language="en")

TODO

• refactoring code
• improve performance