Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.584627
Title: Role of differences in fundamental frequency between competing voices in a reverberant room
Author: Deroche, Mickael
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2009
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
In noisy conversations, listeners can segregate competing voices on the basis of their fundamental frequency (FO). The aim of this thesis was to investigate which mechanisms underlie this FO-segregation ability and whether this ability is affected by reverberation. This work provided evidence for a mechanism, which cancels interfering voices on the basis of their harmonic structure a process termed harmonic cancellation. We developed a paradigm in which listeners had to detect a band of noise masked by a harmonic or inharmonic complex masker (Chapter II). Harmonic cancellation was found to be beneficial up to about 3 kHz, sensitive to a degree of inharmonicity reflected by a peak autocorrelation of 0.9 or less, and to integrate harmonic information over very large bands. In addition to harmonic cancellation, listeners may also use FO as a sequential cue, provided that AFO is sufficiently large (Chapter III), in order to organise the auditory scene in the presence of several talkers a process termed sequential FO-grouping. By manipulating the FO of competing sources heard in anechoic or in reverberant environments, the Speech Reception Threshold (SRT) of a target voice masked by buzz (Chapter IV) or speech (Chapter V) interferers, was elevated when the interferer but not the target, was FO-modulated and especially in reverberation for the buzz interferer. These results were explained in terms of disruption of harmonic cancellation. Moreover, the benefit of an 8 semitone AFO was disrupted by reverberation even for monotonized sources, suggesting that reverberation is also detrimental to sequential FO-grouping. To conclude, the listener's ability to segregate voices by FO relies on the mechanisms of harmonic cancellation and sequential FO-grouping. Both these mechanisms are likely to be disrupted in realistic situations of conversation, i.e. real speech in reverberant rooms.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.584627  DOI: Not available
Share: