bimbu

This is the main scripts for Syllables spotting

Installation

After having cloned the project run:

make install

This will download and install allosaurus which is the main tool used for terms spotting and download the model in Kunwinjku

On Windows

Python 3.6 is needed Install pip. pip is used to install python librairies then you can install the main dependencies:

py -m pip install praatio
py -mpip install pydub
py -m pip install auditok

After that you need to clone Allosaurus project:

git clone https://github.com/xinjli/allosaurus.git

Once done, before the proper installation, go on allosaurus/allosaurus/bin/download_model.py and replace on line 20 the content of url variable by https://github.com/eleferrand/bimbu/blob/main/big_kun.tar.gz you can then go on the main allosaurus folder and run

python setup.py install

and run the following command to unpack the model

python -m allosaurus.bin.download_model

To generate the syllable for a given audio file, run

python transcribe.py <path to your audio>

this will generate a json file with inside the syllable spotted for the given audio file If the audio is two long, it will automatically be broken down into smaller units into the folder splits