Hljóðstafir allows you to automatically sync audio and text highlighting on a sentence level.
Select the language and any experimental features then upload the epub. Hljóðstafir will process the epub file, any issues will be relayed back to the user. Feel free to add tickets to solve any issues you may run into as there are many edge cases that will need to be handled in the future.
- Aeneas may stall without any feedback on longer audio files when running on a Windows OS, therefore it's best to run it on a distro known to support Aeneas. Debian and Ubuntu do not seem to deal with this issue.
Build docker image with Dockerfile
docker build -t hljodstafir .
- Make sure you have python and python cli calls python3 not 2.7
apt-get update && apt-get install -y --no-install-recommends python3 python3-pip python3-dev python-is-python3 curl wget git
- Install Node and yarn
curl -fsSL https://deb.nodesource.com/setup_16.x | bash - && apt-get install -y nodejs build-essential && npm install --global yarn
- Install Aeneas dependencies
wget https://raw.githubusercontent.com/readbeyond/aeneas/master/install_dependencies.sh
bash install_dependencies.sh
- Install python dependencies & aeneas
pip3 install numpy mutagen html5lib
pip3 install aeneas
pip3 install PyMuPDF==1.16.14
- Check whether you installed aeneas correctly
python -m aeneas.diagnostics
- Install node modules
yarn install
- Add required private keys to .env.local
- Run Webapp (production or development)
yarn build && yarn start
yarn dev