Mid-Atlantic Gender Expansive Speech (MAGES) Corpus

What is in this corpus?

This corpus contains the speech of 14 gender expansive (transgender and nonbinary) individuals from the Mid-Atlantic region of the United States. These participants gave consent to have their speech accessible to other speech researchers. Each person recorded 399 sentences, taken from the ModelTalker database (Bunnell et al., 2017), that were chosen to cover the diphones and triphones in American English. Participants recorded remotely in variable recording environments but were required to pass a screening set of ten sentences to make sure the audio quality was acceptable for analysis. Currently, wav files are unprocessed/unfiltered and are sampled at 44.1kHz.

Bunnell, H. T., Lilley, J., & McGrath, K. (2017). The ModelTalker project: A web-based voice banking pipeline for ALS/MND patients. In Proceedings of the 18th Annual Conference of the International Speech Communication Association (INTERSPEECH 2017), 4032–4033

How do I get access to this corpus?

To honor the vulnerability and safety of this community, this corpus is accessible by making an appointment with Maxwell Hope (maxhope@udel.edu) to discuss the intent of its use before access is granted.