How to analyse and manipulate nonlinear phenomena in voice recordings

Area of Science:

• Bioacoustics
• Speech Science
• Auditory Perception

Background:

• Nonlinear phenomena (NLP) in voice production are vital for biological communication.
• Manual annotation of NLP is time-consuming and unreliable.
• Objective acoustic analysis and perceptual studies require robust NLP detection and manipulation methods.

Purpose of the Study:

• To review and present methods for detecting and analyzing nonlinear phenomena (NLP) in acoustic signals.
• To explore techniques for manipulating NLP in audio recordings to test perceptual effects.
• To provide practical tools and algorithms for researchers in bioacoustics and speech science.

Main Methods:

• Utilizing advanced visualization aids like reassigned spectrograms and phasegrams for NLP detection.
• Employing objective acoustic features including general descriptives (e.g., harmonics-to-noise ratio) and nonlinear dynamics statistics (e.g., correlation dimension).
• Implementing algorithms for manipulating NLP, such as adding frequency jumps or subharmonics, and parametric voice synthesis for complex alterations.

Main Results:

• Described objective acoustic measures for quantifying NLP, moving beyond manual annotation.
• Demonstrated feasible methods for manipulating NLP in audio, aiding perceptual research.
• Provided detailed examples with audio files and R code to facilitate practical application.

Conclusions:

• Accurate detection and manipulation of nonlinear phenomena (NLP) are essential for advancing research on vocal communication.
• The reviewed methods and tools offer significant improvements for analyzing vocalizations and their perceptual impact.
• This work supports further investigation into the mechanisms and communicative functions of NLP in vertebrate vocalizations.