Improving multiple-crowd-sourced transcriptions using a speech recogniser | IEEE Conference Publication | IEEE Xplore