Tremor e vibrato humanos: mediação de taxa, profundidade e regularidade no movimento demembros,na voz e no contrabaixo acústico

Abstract

This work invetigated human tremor and vibrato. Algorithms for rate, extent and regularity measurements were evaluated in three signals: (i) tremor in upper limbs; (ii) spoken and sung voice; and (iii) long duration notes in the double bass. The study was carried out by measuring low frequency fluctuations (4-15 Hz) in the fundamental frequency series (F0) and in the signals envelope. Three techniques for amplitude and frequency modulation analysis were studied: analytic signal, Teagers operator, and Fourier analysis. The first two presented difficulties with multicomponent signals such as voice and music, while the Fourier method was more efficient. The mean values measured with Fourier analysis were (a) 4.7 Hz (rate) and 19.83 Hz (extent) for tremor in limbs; (b) 4.8 Hz (rate) and 3.98% (extent) for tremor in spoken voice; (c) 5.7 Hz (rate) and 3.12% (extent) for vibrato in sung voice; and (d) 5.0 Hz(rate) and 0.96% (extent) for vibrato in double bass. The Fourier method was not suitable to track the modulation rate and extent temporal behavior. To overcome this limitation, a technique was proposed based on the autocorrelation function to measure the similarity between F0 cicles over time. Results of application of this method to tremor/vibrato signal suggest that it can be used to measure the fundamental frequency regularity, but it still needs to be better evaluated in future works.