How to extract the most representatives frquencies of a discrete signal

Question

I am given an aperiodic discrete signal (typically music file), and I would like to extract its most "dominant" frequencies over a subset of a signal.

More precisely, I am given N samples of the signal. I have implemented a FFT and them i'm using the FFT values to compute the fourier coefficients. Using their module, I'm basically condluding that the biggest modules are associated with the most representatives frequencies of my signal.

However i have limited knowledge in fourier transform, and so far I have read, the FFT basically gives N points of the fourier transform at frequencies :

fk = k / N

How do I get back to normal frequency scaling? Imagine for instance that one of the most significant frequency is 1200 Hz. How do i find this frequency from the fft (the fft just gives me frequencies between 0 and 1)

I'm rellay confused about that.

Thanks for help.

Answer 1

Let $N$ be the number of samples, $r$ the sample rate (eg. 44000Hz). Now define $T=\frac{N}{r}$. Now your frequencies are given by $f_k =\frac{k}{T}$.