How do you get speech dictated without adding it to a grammar list?

Question

I'm new to Speech Recognition, and I'm working on a project that will receive a command from a recognizable list.

For example, I would say "Play song". The computer would ask the song title, and I can say it. It will then compare my answer to my music library and find it.

I know how to add recognizable grammar to the SpeechRecognizer object, how to make the computer speak, and how to play a song in iTunes. I cannot, however, figure out how to get it to dictate or listen and interpret something that isn't in the grammar list. Is there a method I'm missing? Or not yet been simplified by Microsoft? I have no code to show for this, as I am not even sure how to search for this particular idea.

Of course, I could have the program read my entire library, but that's not an optimal solution considering I have tens of thousands of songs. Thanks in advance!

Answer 1

Speech recognition works by comparing the signal to grammars, so there really isn't a good way to get a result without a grammar. Why not use the music library to generate the grammar list? I'm sure the result will be much more satisfactory. A really general grammar won't enable you to tell whether the user said "Madonna," "My Donna," or "Mad on a" with any sort of reliability. You might want to narrow the grammars further by letting the user choose to say an artist, title or album before they say it.