Re: [w3ctag/design-reviews] Incubation: Web Speech API: On-Device Recognition Quality (Issue #1189)

christianliebel left a comment (w3ctag/design-reviews#1189)

@evanbliu Thank you for your patience. We are aware that on-device speech recognition is already part of the specification. As outlined in https://github.com/w3ctag/design-reviews/issues/1038, we believe that on-device speech recognition is a valuable enhancement to web capabilities.

The additional `quality` field helps determine whether the device can achieve the requested quality level, enabling developers to fall back to alternatives when the quality bar cannot be met. However, introducing recognition quality levels introduces additional permutations that could exacerbate fingerprinting concerns. Previously, an attacker could query for specific languages. Now there's a combination of languages and recognition quality to consider.

That being said, we are closing this design review as **satisfied with concerns**, with the primary concern being the additional fingerprinting vector. We request that you ensure the specification considers robust countermeasures against fingerprinting (see https://github.com/WebAudio/web-speech-api/pull/165), and that the responses to the Security and Privacy questionnaire are added to the explainer.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/w3ctag/design-reviews/issues/1189#issuecomment-4358296757
You are receiving this because you are subscribed to this thread.

Message ID: <w3ctag/design-reviews/issues/1189/4358296757@github.com>