Automated Analysis of Electroglottographic Signal in Adductor Spasmodic Dysphonia
MetadataShow full item record
INTRODUCTION: The human voice can be evaluated by a variety of methods. Electroglottographic (EGG) signal is produced when vocal fold vibrations produce cyclic fluctuation in the electrical impedance across the larynx. The EGG signal thus reflects the degree of contact between the vocal folds during voice production and provides a measure of voice quality based on phonatory physiology. However, the utility of EGG has been limited because existing methods of EGG signal analysis focus on the evaluation of 2-3 parameters in a segment of sustained vowel production, which does not reflect pathologies more apparent in conversational speech. We hypothesize that the EGG signal can capture perceptually relevant information from continuous speech in adductor spasmodic dysphonia (ADSD), an enigmatic speech disorder. OBJECTIVES: 1. To develop an automated computer algorithm to analyze the EGG signal in continuous dysphonic speech. 2. To identify EGG waveform features that correlate with the perceived quality of vocal strain in ADSD. METHODS: A computer program was created and refined in MATLAB to display and analyze EGG data via a graphical user interface (GUI). An automated peak-detection algorithm was developed using the differentiated EGG signal and used to perform simultaneous multi-parameter analysis on the EGG signal from normal speech and speech in patients with ADSD. Between-group comparisons were made using two-tailed Student's t test. Also, intrasubject comparison was made between strained and less-strained syllables in ADSD speech. RESULTS: A program was successfully written to allow the display and automated analysis of EGG data from samples of continuous dysphonic speech. The program was found to generate data with good internal consistency. Application to normal and ADSD subjects showed that the open quotient parameter was able to distinguish between strained and less-strained syllables with statistical significance (p=0.04). DISCUSSION/CONCLUSION: We have developed a method to analyze EGG signal from samples of continuous dysphonic speech. The numerical and graphical data obtained support the utility of EGG as an objective means to clinically highlight the speech differences between normal subjects and subjects with ADSD. Further testing to establish normative values for the analyzed EGG parameters and their subsequent comparison with patient EGG data is required to affirm their utility for routine clinical voice assessment.