Nkululeko: try out / demo a trained model

This is another Nkululeko post that shows you how to demo model that you trained before.

First you need to train a model, e.g. on emodb as shown here
In the ini file you MUST set the parameters

save = True

in the general section and

save = True

in the MODEL section of the configuration file.
Background: we need both the experiment as well as all model files to be saved on disk.

Here is an example script then how to call the demo mode:

python -m nkululeko.demo --config exp_emodb.ini

, if your config file is called exp_emob.ini
Automatically the best performing model will be used.
This will start recording for 3 seconds from your microphone.
if you specify

python -m nkululeko.demo --config exp_emodb.ini --file test.wav

the file test.wav will be predicted, it needs to be in 16 kHz sampling rate and mono channel.
If you specify

python -m nkululeko.demo --config exp_emodb.ini --list my_list.txt --folder data/ravdess/ --outfile my_results.csv

The file my_list.txt will be read and it is expected to contain one file to be predicted per line, e.g.:


The optional argument --folder can be used to specify a parent folder for the input files.

The optional argument --outfile can be used to save the results in a CSV table.

If --list my_file.csv is in CSV format, audformat will be interpreted, but only the index will be used.

E.g. with


only the file column will be used.

Leave a Reply

Your email address will not be published. Required fields are marked *