A set of Python scripts to create GLS packages for use with the Natibo app. The text splitter uses pdftotext which is only available on Linux, sorry Windoze users! Audio splitter should work on Linux/Windoze.
This splits the PDFs into individual sentences and saves all sentences in a tab separated file with the sentence number and various sentence parts (sentence, ipa, romanization, etc.). It relies on pdftotext and isn't available on Windoze, unfortunately. You can try the online version of these scripts if you're on Windoze.
This splits the audio from the GMS files into individual files using the silence as a seperator.
This splits the audio from the GMS files using timings taken from Audacity + a one-line header pointing to the filename
of the audio file. split_gms.py should be used to have automatic conversion. This is just here in case it's important for someone in the future again.
This packages everything together into a zip file and renames it as a .gls file that can be sent to Natibo.
Some things to help clean up the timing files that Audacity spits out.
You will need to create a texts, files and output directory. Put all your PDFs texts, GMS-B files (for example) into files. Run split_text.py followed by split_gms.py and finally create_gls.py.
The results will be in the output directory.
There are various ways of doing this.
Here is an example:
git clone https://github.com/chickendude/GlossikaNativeGLS
cd GlossikaNativeGLS
mkdir texts files output
cp ~/GlossikaCourses/Spanish/ENES-F123-EBK/*.pdf texts
cp ~/GlossikaCourses/Spanish/ENES-F1-GMS/GMS-B/*.mp3 files
python3 ./split_text.py
Checking 'texts/GLOSSIKA-ENES-F2-EBK.pdf'
Checking 'texts/GLOSSIKA-ENES-F1-EBK.pdf'
Checking 'texts/GLOSSIKA-ENES-F3-EBK.pdf'
python3 ./split_gms.py
Analyzing 'files/ENES-F1-GMS-B-0001.mp3'
python3 ./create_gls.py
All code (written by me) is released into the Public Domain.