Text-To-Speech with PyTorch

About • Installation • How To Use • Credits • License

About

This repository contains a model for generating audio from text. The model was trained on LJSpeech dataset.

Installation

Follow these steps to install the project:

0. (Optional) Create and activate new environment using conda.

   # create env
   conda create -n project_env python=PYTHON_VERSION
   
   # activate env
   conda activate project_env

1. Install all required packages

   pip install -r requirements.txt

2. Install speechbrain model for inference

   git clone https://github.com/speechbrain/speechbrain.git
   cd speechbrain
   pip install --editable .
   cd ..
   

3. Load model checkpoint

   mkdir pretrained
   gdown https://drive.google.com/uc?id=16ci_beU3km20_ZFZ4VOFpr4KbIUbKCMi -O pretrained/model_best.pth
   

How To Use

To train a model, run the following command:

python3 train.py -cn=train HYDRA_CONFIG_ARGUMENTS

Where CONFIG_NAME is a config from src/configs and HYDRA_CONFIG_ARGUMENTS are optional arguments.

To run inference (synthesize audio from transcriptions or re-synthesize audio):

# resynthesize - if True, then inputs_dir should contain audio samples, else - text transctriptions
# save_mel - whether to save mel spectrograms to outputs_dir/spectrograms folder
# input_text - if passed then the dataset from inputs_dir will be ignored
# save_name - the save name for audio renerated from input_text

python synthesize.py \
inferencer.inputs_dir="path/to/transcriptions/or/audio/dir" \
inferencer.outputs_dir="path/to/output/dir" \
inferencer.from_pretrained="path/to/model/checkpoint/file" \
inferencer.resynthesize=False \
inferencer.input_text=null \
inferencer.save_name="audio_name" \
inferencer.save_mel=False

Credits

This repository is based on a PyTorch Project Template.

License