Comparison of voice acquisition methodologies in speech research

Comparability of Modern Recording Devices for Speech Analysis: Smartphone, Landline, Laptop, and Hard Disc Recorder

“…This was unexpected given that a previous study [12] reported no significant difference between these systems on the same measures as those examined here ( f0, f0 SD, NHR, number of silences, jitter). The discrepancy between studies may point to the use of less suitable statistical methods for determining reliability and error in the 2008 paper, lending strength to the argument that discrepancies existed between the devices investigated by Vogel and Maruff [12]; however, they were not adequately identified because of insensitive statistical methodologies. Specifically, it is possible to have high RMSE without significant differences if the direction of the individual differences is random.…”

“…Adoption of alternative technologies in voice research has been slow, perhaps due to the requirement for high-quality recording devices needed for some types of analyses (e.g., measures of perturbation) [9,10,11,12]. Historically, only professional-grade data acquisition systems have been accepted as valid and reliable methods for collecting speech data [13].…”

Reliability, Stability, and Sensitivity to Change and Impairment in Acoustic Measures of Timing and Frequency

“…However, the hardware configuration used in the present study has been shown to provide comparable data to higher quality configurations on the acoustic measures selected. 21 Ruling out the recording equipment as a source of variation, another form of technical error may be considered. The analysis program itself is a potential cause of variation; however, several empirical studies have demonstrated the fidelity of Praat 44 and its analytical algorithms.…”

“…For example, in healthy adults, biological variation can reflect diurnal changes in the voice, 19 changes in the voice over time (eg, from aging), 20 or changes in the levels of compliance or motivation in those whose voice is being assessed. In contrast, technological error may arise from the equipment used to acquire data, 21,22 the tools and methods used to analyze data, [23][24][25] or the processes associated with the application of the measure in assessments. 11,26 Technological and biological error occurs in all clinical measurement sciences and can be tolerated provided the error is truly random.…”

Vocal Performance of Group Fitness Instructors Before and After Instruction: Changes in Acoustic Measures and Self-Ratings

“…To optimize the reliability of perturbation and HNR measures, recordings were truncated, leaving 1.5 seconds on each side of the temporal midpoint for analysis. 16 Recordings of connected speech tasks (oral passage reading and conversational speech) were analyzed using measures of F0 and average intensity.…”