I just manually compiled it using javac, then ran it through the jar command. There’s a build.sh script with with source code for compiling it – which I think contains an obfuscation step too.

In principle you could adapt this to work on mobile – assuming you can access the microphone. However this really isn’t a very efficient way of pitch detection. For that you probably need to look at something like the “Fast Fourier Transform”, but that’s quite a bit more complex.

John

I’m new to java, what did you use to compile the jar program?
Also would this work on a mobile, this is my eventual aim.

Thanks

thanks. i am ashamed of myself to have not seen that one

there is an artificial delay that I introduced to stop the app chewing up 100% CPU, so maybe start by removing/reducing that. Look at line 241 in GT.java and you should see a call to Thread.sleep() being passed a value of 250ms (quarter of a second).

John

i am doing a project for my studies and i needed some code to get the frequency out of the mic. so i found your tuner which helped me a lot, but there’s one little thing i have to get rid of though: theres always a delay of about half a second. could you tell me how i could do that. i can send you my project if it helps. thanks in advance

the actual data being read by the call to targetDataLine.read in the code just represents the pressure at a given point in time (so it’s actually the pressure on the microphone at that instance converted to a voltage). The raw data is just a bunch of bytes, so if you us a 16bit sample size (as my code does) you need to join every pair of bytes together to give you one pressure value. That’s the bit with the comment “convert two bytes into single value”.

From there for example, the amplitude of the overall sample can then be worked out by just finding the difference between the highest and lowest values.

Be aware though, this was my first audio programming experience, so you’ll want to double-check everything…

cheers,

John

Do you know how I can extract the amplitudes from the original data? I’m a bit of a novice at signal processing. Any help you can give will be much appreciated.

Thanks,

Mike

well the amplitudes are already present in the original data. The auto-correlation function is trying to take a set of amplitudes and work out their frequency. Analysing the amplitudes should be much more straightforward…

John