MS-Tools

This repository contains the code for analyzing the modulation spectrum.

Prepare

1. Download the audio file and its transcription
2. Convert audio file to .wav format with ffmpeg by setting -ac to 1, -ar to 16000
3. Align and obtain the .TextGrid file with .wav & .txt file by MFA

Modulation spectrogram and spectrum

The modulation spectrogram and spectrum can be calculated by the three function below in the model-test.py script.

# modulation spectrogram
_ = test_spectrogram_generate()
# broadband spectrum
peak_freq = test_spectrum_generate(bandwidth='broad')
# narrowband spectrum
peak_freq = test_spectrum_generate(bandwidth='narrow')

Rate of syllable

Three kinds of rate of syllable can be calculated by the three function below in the model-test.py script.

# syllable rate
syll_rate = test_rate_of_syllable_generate(rate_type='sr')
# articulation rate
syll_rate = test_rate_of_syllable_generate(rate_type='ar')
# syllable mode
syll_rate = test_rate_of_syllable_generate(rate_type='mode')

Correlation

# correlation
demo_of_modeling_relationship_between_syllable_rate_and_modulation_spectrum(data_path='./assets')

Temporal responce function (TRF)

# TRF
demo_of_modeling_relationship_between_syllable_onset_and_broadband_envelope(data_path='./assets')