How accurate is the Audacity spectrum analyzer?

[manisandher]

I've been using Audacity to analyze some measurements of my cartridge (from digitized test tones, etc). I noticed that there was a general downward slope in the 20Hz-20KHz sweep, which surprised me as the cartridge specs indicate it's pretty flat until 24KHz.

 

I then tried a 20Hz-20KHz sweep from the Alan Parson's 'Sound Check 2' test CD. And the slope is there too:

 

 

Playing this track back on my PM2, the meters show a rock steady -15dB output across the whole sweep.

 

So what's going on? Any ideas?

 

Mani.

[Don Hills]

The sweep is a logarithmic sweep. It starts its sweep up slowly and sweeps up faster as the frequency increases. You can hear this, and see it if you click on the down-arrow next to the track title and select "Spectrogram" display. The reason it slopes down in the "Plot Spectrum" display is because the spectrum plotter measures the amount of energy in each "bin". Because the tone spends less time in the higher frequency bins, there's less energy per bin at high frequencies so the displayed level is lower. If you want to see a flat line in the ""Plot Spectrum" display, use a linear sweep so that the time spent in each bin is the same.

 

Generate --> Chirp

Frequency = 20 - 20000

Level = 0.2 - 0.2

Interpolation: Linear (Logarithmic will give the display you showed.)

Duration: 24 seconds

[manisandher]

Thanks Don. That all makes perfect sense.

 

The slope of the 30-second CD sweep appears to be -3dB/octave. The LP sweep's duration is closer to 15 seconds. So would I expect a -6dB/octave slope? Seems a bit steep to me.

 

Cheers,

Mani.

[Don Hills]

Thanks Don. That all makes perfect sense.

 

The slope of the 30-second CD sweep appears to be -3dB/octave. The LP sweep's duration is closer to 15 seconds. So would I expect a -6dB/octave slope? Seems a bit steep to me.

 

Cheers,

Mani.

 

Both slopes should be the same. You can prove it with Audacity: Generate a chirp as before, logarithmic interpolation. Now generate a second chirp with the same parameters as the first except for only half the duration.

They both have the same slope when analysed. Although the tone is sweeping up twice as fast and thus spending only half as much time in each "bin", all "bins" are affected equally.