Article

A Crowdsourcing Survey on Environment Noise Characteristics of the German Crowd-Workers

* Presenting author
Day / Time: 21.03.2019, 16:00-16:20
Room: Saal 3
Typ: Regulärer Vortrag
Session: Lärmbewertung 2
Abstract: Crowdsourcing permits to reach a large pool of users for gathering and annotating data in an efficient and cost effective manner. In a Crowdsourcing context, the users employ their own hardware to execute the tasks from the comfort of their environment. However, there is too little information about the users' background noise and environment characteristics, which is mandatory to judge the validity of the data being collected. Specially, in speech quality assessment and audio related tasks. There have been some attempts to investigate the conditions in which users from English speaking countries, India and Asia conduct crowdsourcing tasks, yet any effort have been made regarding the German crowd-workers. To address this issue, a study in a German-based crowdsourcing platform has been conducted. 220 users participated in total and answered questions regarding the environment in which they normally execute crowdsourcing tasks. Additionally, while conducting the survey, audio and visual data was collected to validate the submitted answers. This paper reports about the environment conditions in which crowdsourcing tasks are being executed. And also the acoustics characteristics of the users' background noises is investigated.