The frequency accuracy depends directly on the accuracy of the sampling clock on the sound card. This is typically a fraction of a Hertz (Hz). You may wish to check your sound card by using a known signal source for comparison.
By default, the analyzer is calibrated to show the relative power levels where 0dB is the strongest possible 8, 16 or 24 bit signal. It is possible to calibrate the amplitude axis by equating a known reference signal level with a measured sampling level; however, the frequency response of your sound card still comes into play (the frequency response of your sound card ultimately determines the ultimate accuracy of the measurement). It is possible to use the mic compensation feature to compensate for the frequency response of your sound card. By measuring the response of a known reference signal (e.g. white noise generator or swept sine measurement) you can determine the frequency response of the input circuitry of your card. This can be placed on the clipboard using the <Edit> <Copy> command, pasted into a text editor and saved as a mic compensation file (you may want to use a spreadsheet to bias the values so that they center around 0 dB).
Octave Scaling Accuracy:
Historically, spectrum analyzers have utilized analog filters for each band and standards (ANSI S1.11-2004) have been developed which specify the performance of these filters . SpectraPLUS-SC uses a Fast Fourier Transform (FFT) to compute the spectral data and then derives the octave data. This algorithm uses the same ISO center frequencies and bandwidths as analog instruments; however, the FFT filtering method produces much steeper "shoulders". These digital filters meet or exceed the performance of traditional analog filters.
The theoretical dynamic range of the system is as follows:
8 bit sampling precision = 48 dB
16 bit sampling precision = 96 dB
24 bit sampling precision = 144 dB
The actual dynamic range achieved is dependent upon the performance of your sound card.
The microphone input on some sound cards utilize Automatic Gain Control (AGC). If this is the case, you will not be able to calibrate the amplitude axis.
See also: Scaling, Settings , Calibration