Riede, Tobias : Vocal changes in animals during disorders

86

Chapter 6. Summary

Vocal changes during disorders in animals were investigated with spectral signal analysis. Two approaches were used depending on if we were dealing with originally harmonic vocalisation or utterances with a considerable amount of noise. First, in harmonic vocalization we determined the amount of nonlinear phenomena. Second, in vocalizations with harmonic and nonharmonic the harmonic-to-noise-ratio was applied. Additionally we were interested in the morphological basis of the vocal changes. In one case study post mortem inspection of the larynx was undertaken to investigate the anatomical basis of the vocal change. In a group of normal dogs the role of the vocal tract in the acoustics of the domestic dog was investigated.

The present work focused on:

1) the quantitative description of subharmonics, biphonation and deterministic chaos (referred to as nonlinear phenomena) in the vocalisation of disordered animals,

2) the possible anatomical basis for the nonlinear phenomena in animal vocalisation,

3) the application of the harmonic-to-noise-ratio to dog barks and its value for vocal change description,

4) the role of the vocal tract in the domestic dog vocalisation.

Normal patients of a veterinary clinic as well as animals from zoological institutions were considered if they expressed vocal changes and if the vocal change could be set into relationship to the disease.

In the first part of the work, three case studies were undertaken. In the first case study a Japanese macaque infant (Macaca fuscata) was investigated expressing a high amount of nonlinear phenomena. On the one hand, this infant showed clear indications of a metabolic disease. The temporal coincidence between the high amount of irregularities (45 %) in the vocalisation and the symptoms of a disease gave the idea of a relationship since in human


87

cry vocalisation the amount of irregularities increases in several diseases. On the other hand, irregularities occurred as common components in the vocal repertoire in nine similar aged healthy infants (on average 13.9 %; min. 3.5 %, max. 41.9 %) but to a lesser extent than in the ill infant. Therefore the hypothesis was proposed that these phenomena are common in mammal vocalisation, but they could be increased in frequency in several diseases.

The second case was a juvenile domestic cat. The conspicuous vocalisation of a three month old cat with craniocerebellar trauma was studied over a period of 8 days in the veterinary clinic. During the first days in the clinic the cat vocalised often and very loud, an effect secondary to a reversible hearing impairment. A temporal coincidence between the improvement of the clinical impression and the change of the vocal utterances were observed. This vocal change could be observed in relevant acoustic parameters but it was seen more clearly in nonlinear phenomena (biphonation, subharmonics and deterministic chaos) in the cat's acoustic repertoire. The vocal change was also recognisable to the unaided ears. The vocal change was probably due to the hyperphonation during the first days in the clinic resulting in an exhaustion of the vocal folds. The voice recovered after 60 days.

The third case deals with a dog-wolf-mix. It was reported to the author that a female dog-wolf mix showed anomalously rough-sounding vocalisation. Spectral analysis of recordings of the vocalisation revealed frequent occurrences of subharmonics, biphonation and chaos. The frequency of nonlinear phenomena in the calls was determined for the female and four additional individuals. It turned out that these phenomena appear, but much less frequently in the repertoire of the four other animals. The larynges of the female and two other individuals were dissected post mortem. There was no apparent asymmetry of the vocal folds but a slight asymmetry of the arytenoid cartilages. The most pronounced difference, however, was an upward extension of both vocal folds of the female. This feature is reminiscent of 'vocal lips' (syn. 'vocal membranes') in some primates and bats. Spectral analysis of the female's voice showed clear similarities with an intensively studied voice of a human who produces biphonation intentionally.

In the second part of this work the harmonic-to-noise-ratio was applied to dog barks In the three case studies mentioned above, only harmonic vocalization was considered. The spectrograms of dog barks show harmonic energy and noise energy to various extent. Vocal changes of the bark, for instance hoarseness, seem to be founded on a shifting of the ratio of the energy between harmonic and noisy elements. Additionally, other authors suggest different communicative relevance according to this energy ratio.


88

Thus, a moving average procedure for calculation of the harmonics-to-noise-ratio (HNR) was applied and tested. First, synthetic sounds with defined HNR confirmed the applicability of the procedure. Second, human subjects evaluated dog barks as predicted by the HNR measure. Third, using the ranking of the animals according their HNR values, it was possible to reproduce the HNR ranking applying multivariate statistics to a parameter set measured on the barks of the same dogs. The results suggest that dysphonia in dog barks can be quantified applying the HNR.

In the third part of this work the role of the vocal tract for the acoustics in the domestic dog vocalisation was investigated. The physical nature of the vocal tract results in the production of formants during vocalisation. In some animals (including humans) receivers can derive information (such as body size) about sender characteristics based on formant characteristics. At least 12,000 years of domestication have resulted in a remarkable diversity of dog breeds. Dogs range in mass from Chihuahuas to Saint Bernards, a 100-fold difference. Because of these significant size differences in breeds, it is not obvious that the correlation between body size, vocal tract length and formant frequencies documented for instance for macaques will hold for domestic dogs of different breeds. The goal of this part of the study was thus to analyse the correlation between vocal tract length, body mass and formant frequencies in the domestic dog.

Lateral radiographs were made of dogs from several breeds ranging in size from a yorkshire terrier (2.5 kg) to a German Shepard (50 kg) and used to measure vocal tract length. In addition we recorded an acoustic signal (growling) from some dogs. Significant correlations were found between vocal tract length, body weight and formant dispersion, suggesting that formant dispersion can deliver information about the body size of the vocalizer. Due to the low correlation between vocal tract length and the first formant we predict a non-uniform vocal tract shape.

The voice/the vocal utterance is an important instrument in different animal communication systems. It can be assumed that clues about the sender‘s state of health are also reflected by vocal changes. Vocal changes have been considered several times as clinical symptom in veterinary medicine, and few data exist on the relevance of vocal changes in animal communication. This study presented data about the quantification of vocal changes in different kinds of vocal utterances in animals. Future studies need to show the communicative relevance of vocal changes, for instance by quantifying them in the natural repertoire of wild animals and applying play back experiments. Simultaneously, anatomical aspects of the source and the vocal tract should be taken into account to gain insights into the generation mechanism of vocal changes.


[Front page] [Acknowledgements] [Attached Audiofiles] [1] [2] [3] [4] [5] [6] [7] [Bibliography] [Declaration]

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiML DTD Version 2.0
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Thu Sep 28 14:01:51 2000