css Copy code

Tacotron 2 Audio Preprocessor

This repository contains a Python script (tacotron2_preprocessor.py) that preprocesses audio files for training a Tacotron 2 text-to-speech model. The script trims silence, normalizes the audio, and saves the processed files to a specified output folder. It's specifically designed to work with .wav files to help create a clean and consistent dataset for Tacotron 2 model training.

Requirements

• Python 3.6 or higher
• Librosa
• SoundFile

Installation

1. Clone this repository to your local machine.

git clone https://github.com/yourusername/tacotron2-audio-preprocessor.git

markdown Copy code

2. Install the required libraries:

pip install librosa soundfile

r Copy code

Usage

1. Update the input_path and output_path variables in the tacotron2_preprocessor.py script to point to your input folder containing the .wav files and the desired output folder for the processed files.

input_path = "path\\to\\your\\input_folder"
output_path = "path\\to\\your\\output_folder"
Run the tacotron2_preprocessor.py script:
Copy code
python tacotron2_preprocessor.py