Relative Salience of Envelope and Fine Structure Cues in Zebra Finch Song
Files
Publication or External Link
Date
Authors
Advisor
Citation
DRUM DOI
Abstract
This dissertation examines the perceptual salience of several acoustic cues in zebra finch song. Birdsong has long served as an animal model of speech development. Both are learned during a sensitive period, and require auditory feedback for learning and maintenance. Zebra finch song is commonly studied due to its stereotyped nature. Song syllables are complex, containing multiple cues that are modulated over millisecond time scales. Using psychoacoustic methods, male zebra finches were tested on discrimination of changes to their own and conspecific songs. Females and budgerigars were also tested, since they have auditory experience with song, but do not sing.
Three types of synthetic songs were created to determine which acoustic cues in song were most salient to birds. Same-seed noise songs were made of syllable envelopes filled with the same piece of random Gaussian noise. This removed spectral structure but kept song envelope cues intact. Random noise songs were made of each syllable envelope filled with a unique piece of noise. This provided more complex fine structure to the same song envelope. Lastly, Schroeder songs were made of Schroeder harmonic waveforms with the same duration as song syllables. In Schroeder waveforms, spectrum and envelope are constant, but phase changes occur across frequencies.
Two types of song changes were tested: single interval duration doublings and single syllable reversals. All birds were much more sensitive to syllable changes than to interval changes. For natural song, there was a duration effect on performance for male zebra finches only. Performance on syllable reversals shorter than 100 milliseconds was positively correlated with syllable duration. In Schroeder song, where only fine temporal structure changes with reversal, all three groups showed a duration effect. Thus, females and budgerigars may focus less on fine structure in natural song than males. In the absence of song spectral structure, birds relied on syllable envelope cues for reversal discrimination. Thus, removal of a single cue from song did not greatly affect reversal discrimination. However, birds performed best when all cues were present. This is reminiscent of human speech, in which multiple redundant cues are used for speech recognition.