Presentation Details
| Assessing the effect of background noise on detection of vocal identity information Parker Major1, Brittany Coppinger1, 2, Grace Smith-Vidaurre1, 2, 3. 1Department of Integrative Biology, Michigan State University, East Lansing, Michigan, USA.2Ecology, Evolution, and Behavior Program, Michigan State University, East Lansing, Michigan, USA.3Department of Computational Mathematics, Science, and Engineering, Michigan State University, East Lansing, Michigan, USA |
Abstract
Many vertebrates can transmit identity information for social recognition through vocalizations. Identity information can be detected by computational tools for empirical analysis, though this may be impacted by background noise. We quantified the effect of background noise on the accuracy of vocal identity information detection. We generated synthetic frequency modulated vocal identity signals with controlled amounts of group identity information using the paRsynth package in R. We designed simulation experiments in which we added three types of background noise (cricket stridulations, white or pink noise) of low or high signal to noise ratio (SNR) over synthetic vocalizations. We used a computational pipeline with spectrographic cross-correlation and unsupervised machine learning to compare identity information detection performance across experimental treatments. We found that the accuracy of information detection was significantly lower for all noise treatments, and the effect of noise on information detection was greater for low SNR treatments. Our results provide guidelines for future work on computational information detection when signals are recorded in noisy environments.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
Content Locked. Log into a registered attendee account to access this presentation.