The Probe Of Nonlinear Dynamic Method On Analysis Of Voice

Part IThe voice effects of sub-glottal pressure in excised canine larynx usingnonlinear dynamics methodObjective: Investigating voice quality between two sub-glottal pressure(SGP) in excised canine larynges, assessing the abilities of describingthe voice signal by nonlinear dynamic methods.Methods: Perturbation measures of jitter percent and shimmer percent,nonlinear dynamic measures of phase space reconstruction and correlationdimension (D2) were applied to analyze the voice from the study includingsix excised canine larynges under2kpa and4kpa sub-glottal pressure.Results: There was a significant difference between2kpa and4kpasub-glottal pressure in D2, jitter and shimmer (P<0.05). However21in60voice signals can not be analyzed by perturbation methods, there wasa significant difference in percent defective of analysis when SGP was4kpa and2kpa (P<0.05).Conclusions: The voice quality when SGP was4kpa maybe better than thatwhen SGP was2kpa, however, we need more experiments if the voice qualitywould be better when SGP was higher. The regulation of voice signal wasrelevant with SGP in the excised canine larynx. Nonlinear dynamic methodswas applied more widely than perturbation methods for voice analysis. Part IINonlinear dynamic analysis in voice signal from patients with vocalpolyps and cystsObjective: To assess and compare the abilities of traditionalperturbation methods and nonlinear dynamic methods to describe voice fromnormal subjects and patients with vocal polyps and vocal cysts.Methods: Perturbation measures of jitter percent and shimmer percent,nonlinear dynamic measures of phase space reconstruction and correlationdimension (D2) were applied to analyze the voice from the study included75human subjects.Results: A significant difference between normal and patients was foundin jitter， shimmer and D2(P<0.05), and no difference between patientswith vocal polyps and with vocal cysts (P>0.05). The phase spacereconstruction is similar between patients with vocal polyps and withvocal cysts, but is more difficulty than normal subjects.D2in the areaunder the ROC curves is less than jitter and shimmer.5patients’ voicesignal cannot be analyzed by perturbation methods, but can be analyzedby nonlinear dynamics methods.Conclusions: Both nonlinear dynamic and traditional perturbationmeasures could differentiate normal and illness voice. Perturbationmethods in the diagnosis of specific are superior to nonlinear dynamics,but have a higher rule of signal and a lower adaptability. Nonlinear dynamic methods can be used as an important supplement to perturbation. Part IIINonlinear Dynamic Analysis of Voices before and after Surgical Excisionof Vocal PolypsObjective: To investigate the effect of nonlinear dynamic analysis bymeans of observing nonlinear dynamic and acoustic perturbation parametersby analyzing voices from vocal polyps before and after surgical excision.Methods: Perturbation measures of jitter percent and shimmer percent,nonlinear dynamic measures of phase space reconstruction and correlationdimension (D2) are employed to analyze voices before and6weeks aftersurgical excision of41human subjects vocal polyps from the study.Results:4patients’ voice signal cannot be analyzed by perturbationmethods, but can be analyzed by nonlinear dynamics methods. Jitter,shimmer and D2of voices from post-surgery were observed to have been lowerthan that pre-surgery, shimmer and D2show a statistically significantdifference(p<0.01), there is not a statistically significantdifference(p=0.054) in jitter.Conclusions: Both nonlinear dynamic and perturbation can evaluate theeffects of surgical excision of vocal polyps; however, perturbationshould be applied to analyze aperiodic voices with caution, nonlineardynamic methods may be useful.