ABSTRACT
Restrictions on face-to-face interactions due to the outbreak of the coronavirus pandemic (COVID-19) in early 2020 have impacted experimental behavioral research. The rapid change from in-person to online data collections has been challenging in many behavioral studies, especially those that require vocal production, and the quality of the remotely collected data needs to be investigated. The current study examines the recording quality and corresponding measures of vocal production accuracy in online and in-person settings using two measurements: harmonic-to-noise ratio (HNR) and fundamental frequency, f0. Participants imitated pitch patterns extracted from recordings of song or speech, either in a laboratory or via an online platform. The results showed that the recordings from the online setting had higher HNR than those from the in-person setting, whereas the pitch imitation accuracy in both settings did not differ. We also report an experiment that simulated differences between the online and in-person settings within participants, focusing on software used, type of microphone, and presence of ambient noise. Pitch accuracy did not differ according to these variables, except ambient noise, whereas HNR again varied across conditions. Based on these results we conclude that measures of pitch accuracy are reliable across these different types of data collection, whereas finer-grained spectral measures like HNR might be affected by various factors.
Acknowledgments
This research was supported in part by NSF Grant BCS-1848930. We thank Tim Pruitt, Emma Greenspon, Fang Liu, Alice Wang, Chen Zhao, David Vollweiler, Kayden Koh, Swathi Das, Jonathan Jun Kit Liow, Anna Gentile, and Kyle Walsh for assistance in stimulus creation; Esther Song, Chantel Fatorma, Kaithlyn Massiah, Thamaraah Bouaz, and Arshpreet Grewal for help in data collection and data processing, as well as Michael Hall and two anonymous reviewers for helpful comments on an earlier version of this manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. Levene’s test for homogeneity of variance was not significant for either of the ANOVAs reported here.
2. Bayesian tests are also robust to deviations from normalcy, and the distribution of pitch deviation scores in do deviate from normalcy according to a Shapiro–Wilk test (p < .001).