The American English SALA-II Data Collection

We discuss the collection of the American English SALA-II speech corpus. We focus on how we designed the prompt sheets to ensure maximum variability and on our strategy for recruiting the required 4000 speakers.We also present results on the effectiveness of the phonetically rich sentence. This paper should benefit others who are interested in using this corpus, or who are planning to collect a speech corpus with a large number of speakers
Published in 2004