Powered by matcha-tts and *retrained montreal forced alignment model (*since matcha-tts is using espeak). A part of the KTH course "Speech Technology". Supervisors: Jens Edlund and Jim O'Regan.
(was tested on Mac Mojave)
1. Download/clone this repo
2. Ensure you have conda and add the conda-forge channel to your Conda configuration
conda config --add channels conda-forge
3. Create virtuatual environment with python>=3.9 and activate it
conda create --name g2p_env python=3.10
4. Activate the environment
conda activate g2p_env
5. Comment out matcha-tts in requirements.txt (because one can install matcha-tts package only using pip)
6. Install requirements using conda
conda install --file requirements.txt
7. Install matcha-tts using pip
pip install matcha-tts
8. Locate your flask app
export FLASK_APP="YOU_ABSOLUTE_PATH/g2p_correction/g2p_correction.py"
8a. If you are having problems with locating your flask app within your venv, try this:
export FLASK_ENV=development
9. Set up your database
cd YOU_ABSOLUTE_PATH/g2p_corrections
flask db init
flask db migrate -m "Initial migration"
flask db upgrade
flask run
1. Download/clone the repository
2. Have your Docker app open
3. If you don't have an ubuntu base image amongst your docker images, run this
docker pull ubuntu:22.04
4. Being inside the g2p_correction directory, build the Docker image:
docker build -t g2p_correction:latest .
docker run --name g2p_correction_container -p 5000:5000 --rm g2p_correction:latest