...
Hello, everyone!
I am a student trying to make a Do-It-Yourself device regarding to sound recognition. Not too much precise and complex as speech recognition, I just want to adopt and use some simple recognition library for my small app.
What I want to do is, in other words, is a sound recognition with "wide-range". i.e. if there is a sound "Boom! Boom!" then I don't really have to recognize that such as, "Okay, that is a certain engine sound from Chevrolet Corvette Z06. Nice machine, huh?" (X) but rather to recognize that as something like, "Car. Coming to me. Loud and Noisy." (O) This is enough.
or, if there is a sound "Ka-boom!" "Bang!", I don't have to recognize it as, "Firework play at the north bridge. In this winter? Maybe some retarded are out there." (X) but, "Fire boom. I would never go there. Strange." (O) this is enough. ... and so on. That is, I want to 'categorize' sounds with the anonymous library and use it as a useful information for guidelines given by my app.
Is there any helpful information, or recommendable libraries I can refer to? I would be awaiting your great advice. Thank you.
Regards,
J. Park
if you're open for a C++ library (not really C so) :
But you'll need to developp yourself the reco, it's done by using a base of samples which are labelled as the noise they do. It's a long project, do you are familiar with signal processing as Fourier,DTW, etc ? There are standard process to do this.