Although you're not going to replicate the voice recognition system in that popular smartphone, with µSpeech you can have up to ten words that are recognised in order to execute various commands and so on. Unlike other systems µSpeech doesn't require any training, instead it uses a mathematical algorithm to analyse the sounds' numerical representation and return what was said. Hopefully! According to the website the accuracy is up to 80% for words, so again considering the level of hardware it's a great effort.
For more information, the library download and tutorial visit the project page. And for more, we're on twitter and Google+, so follow us for news and product updates as well.
The MIC: contains an integrated dual signal amplifier converts the sound to separate channels for pulse / frequency measurement, and sound volume (pressure) level. Designed to connect straight to an Arduino compatible microcontroller, Analog to Digital converter, or many other circuits. For more information and to order click here.