Spatiotemporal droplet dispersion measurements demonstrate face masks reduce risks from singing

COVID-19 has restricted singing in communal worship. We sought to understand variations in droplet transmission and the impact of wearing face masks. Using rapid laser planar imaging, we measured droplets while participants exhaled, said 'hello' or 'snake', sang a note or 'Happy Birthday', with and without surgical face masks. We measured mean velocity magnitude (MVM), time averaged droplet number (TADN) and maximum droplet number (MDN). Multilevel regression models were used. In 20 participants, sound intensity was 71 dB for speaking and 85 dB for singing (p < 0.001). MVM was similar for all tasks with no clear hierarchy between vocal tasks or people and > 85% reduction wearing face masks. Droplet transmission varied widely, particularly for singing. Masks decreased TADN by 99% (p < 0.001) and MDN by 98% (p < 0.001) for singing and 86-97% for other tasks. Masks reduced variance by up to 48%. When wearing a mask, neither singing task transmitted more droplets than exhaling. In conclusion, wide variation exists for droplet production. This significantly reduced when wearing face masks. Singing during religious worship wearing a face mask appears as safe as exhaling or talking. This has implications for UK public health guidance during the COVID-19 pandemic.