What does a FFT plot show?

I have plotted a FFT of a sound me saying Jump. Can Someone help me please. 1.What useful information I can find from this graph? 2.Does the point I put in the graph is the frequency at which I said the word jump?

Answer 1

[dpb]

Not sure this one shows you much of anything...you've plotted the double-sided spectrum which isn't generally of much value; might as well just use the DC-based to Nyquist/2 positive frequency beginning at N/2+1 for the location.

But, the sharp peaks at harmonics looks to me like you most likely badly clipped the input and you've mostly got the spectrum of a square (or nearly square) wave...

[JOJO]

1.-I use Darryl Morrell youtube video on "Graphing a Frequency Spectrum with Matlab" to plot the frequency domain plot, it is not good this way? The code I used is at the end of his video shown 2.-how do I make it better?

[Image Analyst]

You forgot to give the youtube link. Why not make it easy for people to help you, not hard? If you force someone to look up something like that, they may just not bother.

[dpb]

There's nothing really wrong w/ the plot, per se, it's the underlying data I question. The peaks look too suspiciously like a square wave to be a (unpolluted) voice recording; hence I suspect the sampled signal was severely clipped or otherwise contaminated.

The comment on the plot is only that there's really no reason for the negative frequency for visualization but that's a nit...the other problem is still there...

[JOJO]

-Image Analyst when I put links Matlab delete them. http://www.youtube.com/watch?v=dwzQnbeKnQg

-When I recorded the sound I talked loud?This caused the clip? -There is a way to make the data better use a filter or something?

[dpb]

...sound I talked loud?This caused the clip?

Maybe, maybe there's some other contamination source. How did you record the sound, how did you process it, did you plot it to see what the time record looks like, did you ensure you only analyzed a section that contains only the desired sound snippet, is there any antialiasing protection, what's the sampling rate, duration, etc., etc., etc., ...???

We have no way of knowing any of this and more...

The rolloff to DC is somewhat suspicious as well...