Comparison of nonlinear dynamic methods and perturbation methods for voice analysis

Abstract

Nonlinear dynamic methods and perturbation methods are compared in terms of the effects of signal length, sampling rate, and noise. Results of theoretical and experimental studies quantitatively show that measurements representing frequency and amplitude perturbations are not applicable to chaotic signals because of difficulties in pitch tracking and sensitivity to initial state differences. Perturbation analyses are only reliable when applied to nearly periodic voice samples of sufficiently long signal lengths that were obtained at high sampling rates and low noise levels. In contrast, nonlinear dynamic methods, such as correlation dimension, allow the quantification of chaotic time series. Additionally, the correlation dimension method presents a more stable analysis of nearly periodic voice samples for shorter signal lengths, lower sampling rates, and higher noise levels. The correlation dimension method avoids some of the methodological issues associated with perturbation methods, and may potentially improve the ability for real time analysis as well as reduce costs in experimental designs for objectively assessing voice disorders.

Updating signal typing in voice: Addition of type 4 signals

“…Detailed descriptions of this method can be found in numerous publications and therefore will be described only briefly here ͑Jiang and Zhang, 2002;Zhang et al, 2005aZhang et al, , 2005bZhang and Jiang, 2008͒. D 2 was used to quantify the nonlinearity of the voice signals in this study.…”

“…Unlike perturbation analysis which is dependent on the accurate determination of fundamental frequency, nonlinear analysis can be completed without establishing a fundamental frequency ͑Ma and Yiu, 2005;Zhang et al, 2005b͒. Accordingly, nonlinear dynamic analyses, such as correlation dimension ͑D 2 ͒, have been shown to be reliable for use on all three of Titze's ͑1995͒ voice types ͑Zhang et al., 2005b;Little et al, 2007;Zhang and Jiang, 2008͒. Despite the improvements afforded by nonlinear analysis, many of its parameters are negatively impacted by infinite dimensionality ͑Little et al, 2007͒.…”

Asymmetric spatiotemporal chaos induced by a polypoid mass in the excised larynx

“…Recent studies have shown that the dimension estimate can be obtained when the digitized noise of a voice signal is as high as 2%. 32 Thus, nonlinear dynamic analyses can be applicable for high-speed image data with a high image resolution. Figure 6͑b͒ shows the dependences of the estimated dimensions on m, where the curves from bottom to top correspond to a 1 ͑1͒ ͑t͒, a 1 ͑2͒ ͑t͒, a 1 ͑3͒ ͑t͒, a 1 ͑4͒ ͑t͒, and white Gaussian noise, respectively.…”

Effects of Melody and Technique on Acoustical and Musical Features of Western Operatic Singing Voices

“…However, these findings should be compared with perceptual judgments. In fact, parameters such as jitter, shimmer, and SNR have been shown to be relevant for complex acoustical signals, [35][36][37] but the analytical tools may be limited; for example, a strong perturbation may lead to a failed F 0 extraction in perturbation analysis and thus provide irrelevant information. 53 Regarding the significant correlations between VE and the perturbation parameters, our results suggest the need to control for VE in future research on the aperiodicity of operatic singing voices.…”

“…34 Despite the limitations regarding the control of the musical samples, these results suggest that perturbation analysis (ie, parameters such as jitter, shimmer, and signal-to-noise ratio [SNR]), which is commonly used for evaluating voice disorders, 3 is relevant for operatic voices. [35][36][37] As well as investigating the vocal and quality parameters of the operatic singing technique (ie, VR and VE, spectral distribution, and vocal perturbation), this study examined some musical characteristics such as the loudness of the voice and fundamental frequency (F 0 ) of the starting note and tempo. In a comparison of the performance of a popular song with and without an operatic singing technique, a particularly slow tempo was observed when the song was performed with an operatic singing technique.…”