This library provides a way to design listening tests in Python.
The library contains several functions for audio processing, such as convolution or different types of normalization, including loudness normalization using the mosqito library.
Only ABX test protocol is currently supported, but more protocols will be added in the future.
It is currently not possible to install this library using pip or conda, please use the latest released package instead and install using pip locally.
API documentation can be found here.
The following script implements a basic ABX listening test using the listeningpy library.
More examples are to be added in the future to the examples folder.
import customtkinter as ctk
import logging
import argparse
from pandas import DataFrame
from time import strftime
from listeningpy.gui import abx
from listeningpy.gui.gui import InitFrame
from listeningpy.set_preparation import abx_combination, randomization
from listeningpy.processing import straight
logging.basicConfig(datefmt='%H:%M:%S',
level=logging.INFO)
class ABXTest(ctk.CTk):
"""Main class for the ABX test GUI."""
def __init__(self, **kwargs):
ctk.CTk.__init__(self, **kwargs)
self._frame = None
def switch_frame(self, new_frame):
"""Destroys current frame and replaces it with a new one."""
if self._frame is not None:
self._frame.destroy()
self._frame = new_frame
self._frame.place(relx=0.5, rely=0.5, anchor=ctk.CENTER)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('-sounds', default='stimuli', help='Directory path for sounds')
args = parser.parse_args()
sounds_dir = args.sounds
# Read files in the specified folder and generate combinations for ABX test
df = abx_combination(sounds_dir, '.', constant_reference=False)
df = randomization(df)
# Create root window and GUI
root = ABXTest()
root.title('ABX test v0.0.1')
test = abx.Abx(root,
df,
processing_func=straight,
)
init = InitFrame(root, test)
root._frame = init
init.place(relx=0.5, rely=0.5, anchor=ctk.CENTER)
root.mainloop()
init_dict = {
'first_name': init.first_name,
'second_name': init.second_name,
'date_birth': init.date_of_birth,
'gender': init.gender,
'hearing_impaired': init.hear_impaired
}
timestamp = strftime("%y-%m-%d_%H-%M")
# Save user's information and test results to CSV files
DataFrame(init_dict, index=[0]).to_csv(
f'{timestamp}_{init.first_name}_'+
f'{init.second_name}_info.csv'
)
test.combinations.to_csv(
f'{timestamp}_{init.first_name}_'+
f'{init.second_name}_results.csv'
)This library was created thanks to the FAST-J-23-8284 project.
Special thanks to Prof. Monika Rychtarikova for her help with the theory and design of the listening tests and also to Prof. Christ Glorieux, Yannick Sluyts and Dominika Hudokova for their valuable feedback.
Github Copilot was used to generate parts of the documentation and code.
-
PhD student at Brno University of Technology and KU Leuven
Pull requests are welcome. For any changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.