Microsoft has started rolling out some important updates to give users more control over how their voice data is used to improve its speech recognition technology. The new setting will allow users to choose if the company can use their voice recordings to enhance the performance of its voice-enabled products and services.
Microsoft’s data collection practices came under the radar back in 2019. A report revealed that the company was using human contractors to listen to voice conversations done with Cortana and Skype Translator. Last year, the software giant officially acknowledged it on its privacy policy and product support pages and stopped storing voice clips processed by its speech recognition technologies over privacy concerns.
The latest announcement comes in response to these privacy concerns about the use of audio content. From now on, the software giant will no longer force users to share voice clips, and it will roll out a new update for various products and services that use the speech recognition technology. The list includes Microsoft Translator, Windows, Skype voice translation, HoloLens, Cortana, SwiftKey, and Mixed Reality.
In the blog post, the company now explains that it’s introducing an opt-in program for customers to submit voice recordings to improve the performance of its services. If people choose to opt-in, the human contractors and Microsoft employees may transcribe the voice content, and it will be retained for up to two years. Those customers who don’t want to participate in this program will still be able to use the company’s services.
“While Microsoft employees and contractors will only listen to voice clips with user permission, the company may continue to access information associated with user voice activity, such as the transcriptions automatically generated during user interactions with speech recognition AI. The details of how that works are described in the terms of use for individual Microsoft products and services, the company said,” the blog post reads.
Microsoft noted that personal data and personal identifiers (including email addresses, telephone numbers, and social security numbers) will be removed from voice clips before processing them. The company added that human contractors don’t use audio data collected from certain enterprise-focused speech recognition features.
