In a matter of six months, the image of people at stores, walking college campuses, and eating out has been altered by a simple, but significant change: a mask. Researchers at the University of Illinois wanted to know if, or how, the addition of a face mask could affect automatic speech recognition (ASR).
The study consisted of the acoustic evaluation of 22 types of masks. The types of masks involved in the study included: polypropylene surgical mask, N95 and KN95 respirators, six cloth masks made from different fabrics, two cloth masks with transparent windows, and a plastic shield. The study, which was performed in an “acoustically treated lab” found that (drum roll) the best type of mask for acoustic performance was the 100% cloth masks. The worst? The masks with transparent windows.
Summary of the study? It was determined that, “most masks had “little effect” on lapel microphones, suggesting existing systems might be able to recognize muffled speech without issue.”
The Venture Beat article referenced asked for comments from both Google and Amazon. Their responses were that they haven’t seen an issue or shift on their automatic speech recognition systems. Good news for the Google Home and Amazon Echo fans out there. The changes that COVID-19 brings to the voice world will continue to be assessed and monitored as the industry works creatively to bring solutions in unique ways.
To read more in depth on the study, click here.
