I'm working with the HTML 5 speech input API and I want to let the server know which answers it can expect to be returned from the speech input. Is there a way to set a list of possible inputs?
In Google Chrome you can not use grammars yet, overall they decided to use free-form recognition only. Relevant question is Grammar in Google speech API.
Grammars are supported in Firefox Web Speech API, but the feature is experimental.
If your browser supports HTML5 audio, you might want to try to recognize words from your grammar using Pocketsphinx.js library, it provides javascript-based recognizer which should be very good for grammars.
I assume your a talking about the Speech Input API Specification as defined by the W3C. According to this specification grammars are supported, which is the method used by speech recognition systems to restrict possible inputs. Look for the section titled "Speech recognition grammars". The problem will be how well it is supported in the particular browser that is being used, which is the case for an HTML 5 specific function since it is not a completed standard yet. Chrome seems to be the most up to date with the standard. Checkout this blog post from the Chrome team on implementing the Speech Input API.
来源：https://stackoverflow.com/questions/8683772/is-there-a-way-to-use-a-grammar-with-the-html-5-speech-input-api