Voicing Erasure project
In March 2020, Algorithmic Justice League released a project titled Voicing Erasure Project[1][2][3]. A project performed by various women and led by Allison Koenecke. These Automated speech recognition (ASR) systems are known as sophisticated machines that identify and convert spoken language into text. In this project they address the notion of racial bias in speech recognition algorithms. They also examine various ASR systems that are developed by Amazon, Apple, Google, IBM, and Microsoft - through the process, they transcribe the interviews they conducted with 42 white individuals and 73 black individuals in order to demonstrate the racial disparities based on the performance of five commercial ASR systems. The end results determined racial disparities in all five commercial systems, with an average word error rate of 0.35 for black speakers, in comparison to 0.19 for white speakers. Therefore depicting machine learning systems as a database heavily relying on English spoken by white Americans.
References
- ↑ "Voicing Erasure". www.ajl.org. Retrieved 2021-11-16.
- ↑ Koenecke, Allison; Nam, Andrew; Lake, Emily; Nudell, Joe; Quartey, Minnie; Mengesha, Zion; Toups, Connor; Rickford, John R.; Jurafsky, Dan; Goel, Sharad (2020-04-07). "Racial disparities in automated speech recognition". Proceedings of the National Academy of Sciences. 117 (14): 7684–7689. doi:10.1073/pnas.1915768117. ISSN 0027-8424. PMID 32205437 Check
|pmid=value (help). - ↑ University, Stanford (2020-03-23). "Automated speech recognition less accurate for blacks". Stanford News. Retrieved 2021-11-16.
This article "Voicing Erasure project" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Voicing Erasure project. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
