[Jimmmeh]

Hey guys, for my old Series 60 2nd Edition phone I had a great little Guitar Tuning app which could work out the frequency of a note captured live from the phone microphone.

I imagine this used some sort of FFT (Fast Fourier Transform) on the incoming wave data to detect the dominant frequency. I was wondering if it would be possible to re-create in J2ME using the MMAPI.

I know a live stream wouldn't be able to be processed but do you think it would be possible to have a thread capturing 1/4 second bursts, writing to a byte array which could then have the PCM data extracted from it and run through some FFT algorithm? If so does anyone know of a very simple version of that algorithm, written in Java, that only makes use of functions available in J2ME (or one which could be translated easily).

I know J2ME apps are a lot slower than native Symbian C++ but if a native app could perform FFT translations "live" then surely Java would be capable of processing short bursts of samples?

FFT is a bit over my head really and I know there are C++ libraries such as FFTW available but I just want a small, simple algorithm to just let me work out the dominant frequency of the sample sound.

If anyone can help at all I would be extremely greatful as it would be great to have a J2ME guitar tuning app to easily take around. Thanks in advance to anyone who can provide me with any help here!

[redlex]

Hi Jimmmeh
I know of no reason why this shouldn't work! As you note, it may be better to analyse a half second sample, rather than attempt live FFT.

As for code, googling pulls up some useful looking stuff, e.g.:

http://www.developer.com/java/other/article.php/3457251
http://www.cs.princeton.edu/introcs/.../FFT.java.html
http://www.dsptutor.freeuk.com/analy...mAnalyser.html

[Jimmmeh]

That FFT algorithm looks good but I am not entirely sure how to go from the data in the PCM wav file to the inputs to the algorithm. Like, I am not sure how i need to transform the data in the wav file to be used as an input in the FFT algorithm.

There are lots of articles on FFT but none that explicitly talk about moving from the data in a wav file to a frequency spectrum. Anyone else have any input on that?

Cheers again and thanks very much for your help redlex!