Consider redesigning the Web Speech API with Promises & ReadableStreams

[evanbliu]

The Web Speech API's reliance on events and callbacks hinders its usability in modern web development. Adopting Promises would streamline asynchronous operations like speech recognition and synthesis, improving code readability and error handling. Modernizing the API with Promises and ReadableStreams aligns it with current web standards, promotes cleaner code, and unlocks new potential for web-based speech applications.

While the existing Web Speech API functionality will remain to maintain backwards compatibility, we should consider if the pros/cons warrant the development of a new, modernized version of the Web Speech API.

https://www.w3.org/TR/design-principles/#promises
https://www.w3.org/2001/tag/doc/promises-guide

[bkardell]

Yes, please - I have been asking for this since 2017. I would love to maybe be involved in this... Back then I wrote several pieces on this and had sketched a kind of a shape for Speech To Text in https://bkardell.com/blog/Voice-Recognition-Thoughts.html, and Text to Speech in https://bkardell.com/blog/Basic-Voice-Speaker.html ... I haven't looked at them in some time to see if I still agree with all of my observations but I'd definitely still be interested in addressing some of these. I recently saw that .start(MediaStreamTrack) shipped in chrome... it seemed like it might have some of those issues I remembered so ran over to try throwing an mp3 at it only to be sad that it didn't seem to transcribe more than the first few seconds..