Speech-to-text fairness information
Microsoft performed a fairness assessment of our Speech to text API using a benchmark dataset for different demographic groups based on age and gender.
Introduction
At Microsoft, we strive to empower every person on the plant to achieve more. An essential part of this goal is working to create technologies and products that are fair and inclusive. One dimension of this goal is to consider and evaluate how well the system performs for different groups of people. Research has shown that without conscious effort focused on improving performance for all groups, it is often possible for the performance of a system to vary across groups based on factors such as race, ethnicity, region, gender, and age.
As a part of Microsoft’s commitment to this fairness goal, the speech team performed assessments of the Speech-to-text capability using benchmark datasets for different age and gender groups. The speech team collaborated directly with responsible AI and external sociolinguistic experts to develop the data collection and evaluation approaches outlined in this assessment.
Microsoft and the speech team are continuing to work with experts to improve their fairness approach and publish more comprehensive fairness results for additional speech services.
How does this impact you?
Microsoft’s internal fairness testing aims to provide customers and partners with the necessary transparency to assess the limitations and appropriateness of the technology for their intended use. We encourage you to review the following resources:
Explore Speech-to-text fairness assessment
![](/assets/rai/images/bar-chart.png)
Explore how we assessed the fairness of the Speech-to-text technology by conducting different types of performance evaluations.
Perform your own assessment and review additional resources
![](/assets/rai/images/speakers.png)
Review next steps to perform your own assessment, understand limitations and trade-offs of the service, and learn more about Microsoft’s Responsible AI strategy.