Country report — Singapore

• To collect speech data over mobile devices primarily for the development and evaluation of automatic speaker recognition systems. • Through a mobile crowd-sourcing platform consisting of a front-end application running on mobile devices communicating with a centralized web server at the back-end. • Mobile crowd-sourcing: — Exploiting mobility and realistic speech capture in the field — Uses Internet, instead of telephone, which gives the benefit of wide accessibility beyond specific regions at a much lower cost • Speech recordings were collected by having speakers read text prompts displayed on the screen of the mobile devices. • Large number of sessions were collected from each speaker over a long time span, typically one session per week over one year period. The corpus could therefore include rich inter-speaker and intra-speaker variations.