Collecting Speech from Crowds
Ian McGraw · 2013
This chapter gives the reader the basic tools necessary to perform their own speech collection tasks. The first section begins by giving a short history of speech collection, describing how the field has progressed to date. It asserts that web-based speech collection, while not yet well explored, has the potential to vastly simplify audio collection efforts. It describes the current technology for microphone-enabled websites in second section and points to future technologies, such as HTML5, which promise to simplify the task even further. In third and fourth sections, The chapter provides an example audio recorder and server. The fifth section outlines the steps necessary to deploy this audio collection interface to mechanical turk (MTurk), to crowdsource the collection of speech. The sixth section talks about the usage of the platform purely for payment. The chapter concludes with a look to the future in the seventh section.