Using Voice Recognition Software to Collect Fisheries Biodata: A Proof of Concept

By Shawn P. Sitar and Daniel J. Traynor 

Biological data collection and processing involves numerous steps, is time consuming, and is error prone. We developed the Voice Data Recording System (VDRS) that eliminates multiple steps in the traditional process by integrating voice recognition software with a normalized, relational database to capture data onto a laptop computer. We compared traditional and VDRS methods and report efficiency and error rates based on Lake Superior Lake Trout Salvelinus namaycush survey data collected during 2000–2016. Efficiency was measured by data collection rate and processing duration. Data collection rates were not different between methods. Processing duration for the traditional method averaged 200 staff hours per year, whereas the VDRS has no data processing. Error rates measured for the traditional data method averaged 5% and the VDRS had 0% error. We found that the VDRS is more efficient and less error prone than the traditional method of data collection and provides substantial cost and time savings.