How to get the frequency plot spectrum of a wav file?

vvn2023 · 06-09-2016, 05:47 AM

Hi,

I have a wav file with sample rate 48kHz.
In that file I kept a sine tone with 1kHz frequency.

I want to check these using sox or any other utility that was available in the linux.

I used sox file.wav -n stat -freq but I got lots of data which I dont know how to analyze.

Can anyone help me in this

Keruskerfuerst · 06-09-2016, 08:04 AM

You need a fast fourier transformation.

vvn2023 · 06-09-2016, 08:07 AM

How to get that?

Keruskerfuerst · 06-09-2016, 02:34 PM

AVX or AVX2 of modern CPU is FFT.

vvn2023 · 06-10-2016, 12:10 AM

When I used this command
sox $1 -n stat -freq 2>&1 | sed -n -e :a -e '1,15!{P;N;D;};N;ba' | gnuplot -p -e 'set logscale x; plot "-" with linesp'

I am getting the plot specturm.
But I want that as a plain text in terminal output.

Is that possible?

keefaz · 06-10-2016, 07:54 AM

Just remove the gnuplot & plot commands from the pipe?

vvn2023 · 06-10-2016, 08:47 AM

Removing wont solve. It gives a lots of frequency and other output.
But I need the peak frequency

jefro · 06-11-2016, 07:17 PM

Wow, didn't know it could do so much. http://sox.sourceforge.net/sox.pdf

Can't find where sox can find maximum frequency however.

See this maybe. http://superuser.com/questions/43626...-an-audio-file

If you wanted to look at it visually you might be able to use Audacity. https://pantherfile.uwm.edu/type/www...isAudacity.htm

vvn2023 · 06-12-2016, 10:39 PM

Hi,

I am unable to comment on that link.
Could you please let me know which language it is.
I am using bash for doing my stuff.
How to use that code here.?
Or do i need to change my language?

vvn2023 · 06-13-2016, 06:02 AM

I used something in python like
code:

import scikits.audiolab, scipy
import pylab
x, fs, nbits = scikits.audiolab.wavread("capture_smoke_rec_16000_16.wav");
X = scipy.fft(x)
print X

I got this output

[[-0.13085938+0.j 0.00000000+0.j]
[-0.76538086+0.j 0.00000000+0.j]
[-1.41424561+0.j 0.00000000+0.j]
...,
[ 1.84771729+0.j 0.00000000+0.j]
[ 1.41418457+0.j 0.00000000+0.j]
[ 0.76531982+0.j 0.00000000+0.j]]

I dont know how to analyze this one.
I thought I will get 1kHz as the peak frequency.
How to get that