This was made using the Web Speech API and a matching system I came up with that uses Levenshtein distances and tokenized string matching. I wrote up some demo code here: https://github.com/benwasser/WebSpeechAPI
This is cool and all, but it's another Chrome only site. You might as well have another desktop app that I have to download. Why isn't there support for other major, Modern browsers such as FF, Opera and IE11?
Sorry about that! It uses the nascent Web Speech API which seems to rely on external audio processing (i.e. how Siri works). Google has such capabilities so that's why it currently works with Chrome (although I think it works in Safari, as well).
This is brilliant! One feature request though: It'd be cool if the microphone icon toggled the voice recognition, rather than simply enabling it. I couldn't figure out if there was a way to disable it (besides closing the tab and reopening it).
That's not the intended behavior. I developed it on Chrome for Mac, so it should just listen continuously. If you have a second, could you let me know what OS and browser versions you're on?
Also, I agree that it could be a cool service, but my wrapper is relatively light compared to the work being done by Google, so I'm guessing you'll start to see a lot of competition in this area.
11 comments
[ 2.5 ms ] story [ 18.7 ms ] threadHow many predefined answers do you have?
It currently has 64 predefined answers with multiple triggers linked to each one.
great idea btw. I wondered instantly if this could be turned into a service.
Also, I agree that it could be a cool service, but my wrapper is relatively light compared to the work being done by Google, so I'm guessing you'll start to see a lot of competition in this area.
French power I guess