Well, if you sample the signal from the guitar fast enough you should come up with the wave form. I don't know what the frequency of the guitar is but I think human hearing range is up to about 22 kHz. In that case you need to sample at least 44 kHz fast which is possible with the 16 MHz ATmega168 chip. So what you could do is save the last sample. Then get the current sample and subtract the last sample from the current sample to find the delta. Err, also save the last delta. If you notice the "delta" change signs then you know that the sinusoidal wave has just passed a peak or trough? If that's the case then you can take the sample there as the amplitude of the wave. However, this assumes a sinusoidal wave which I'm guessing might not be the case. See if you can just log the data and plot it to see what the wave looks like. If you're crazy enough you could run a discrete FFT... just an idea. Or if anything you could just keep a running max or min of the samples in the past 100 samples or something and use that as your amplitude. Bah, I just realized that the Arduino can't really pick up the negative voltages since the ADC goes from 0V to 5V. Well, you could still do a similar analysis I think.