How can I capture speech in a browser?

I would like to build a reference website that includes flash-cards for people learning new things. The simple idea is to display a card for a time and have the user say the meaning of the card.

I would then like to access the user's microphone, capture their response, and pass it to the back-end for speech recognition.

I only want to know how to capture the speech, the rest I can easily do myself.