Question

So I have an array of transformed complex numbers. I do not know what do with the returned data. I've tried calculating the magnitude each element, designating the one with the highest element as the correct frequency, but found that I get weird numbers like 5,000hz often. Is there a better operation I can perform on the returned data to determine pitch?

Note: I want to use the pitch as part of Voice Activity Detection Algorithm.

Was it helpful?

Solution

There are many different algorithms for pitch detection - I recommend you try using the Harmonic Product Spectrum method, since you have already embarked down the FFT-based route, and it's fairly simple to implement one you have a power spectrum.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top