CLEESE: An open-source audio-transformation toolbox for data-driven experiments in speech and music cognition

Area of Science:

• Auditory cognitive neuroscience
• Psychoacoustics
• Speech and music processing

Background:

• Visual social cognition advanced with graphics tools for face processing.
• Auditory studies lack tools for higher-level cognitive processing (e.g., words, music).
• Reverse correlation is limited to spectral/spectro-temporal properties, not complex auditory cognition.

Purpose of the Study:

• Introduce CLEESE, an open-source audio transformation toolbox.
• Enable systematic randomization of prosody and melody in speech and music.
• Provide tools for studying higher-level auditory cognitive processes.

Main Methods:

• CLEESE cuts audio into small time segments.
• Applies random parametric transformations (pitch, duration, amplitude) to segments.
• Utilizes a Python phase-vocoder implementation for audio manipulation.

Main Results:

• Demonstrated CLEESE's capability to generate novel auditory stimuli.
• Presented applications in studying intonation processing (interrogative vs. declarative speech).
• Showcased application in studying rhythm processing of sung melodies.

Conclusions:

• CLEESE offers a novel approach to manipulating auditory stimuli.
• Facilitates research into complex auditory perception and cognition.
• Opens new avenues for data-driven studies in auditory domains.