Fastest way to conduct frequency sweep for audio signal

[Quintin]

I need to find the frequency dependent complex transfer function between a reference audio signal and a PCM signal from a DUT. 

My question is what is the fastest/best way to send the analog signal if I want it to cover 20Hz to 20kHz range?  Sweep, finite # of points, random noise, mulitone, pseudo random noise, other?

[Quintin]

My description wasnt that clear:

My DUT is a microphone and I want to test the frequency response between the DUT mic and a calibrated reference microphone.  Now to do this I want to output a analog signal from my PC via a loudspeaker to be the sound source that the microphones will be reading. 

My question is what is the best/fastest way to generate this analog sound source signal if after the test I will need to extract the magnitude and phase of the freq. response at multiple frequencies?

My entire test time needs to be less than 1 second.

[johnsold]

Broadband noise will excite the entire spectrum simultaneously. A continuous tone at a single frequency is easy conceptually and does not require you to measure the acquired frequency since is is known from the excitation. To measure the amplitude and phase at one frequency requires several cycles of signal, depending on the method of measurement. FFT based methods, for example, are not happy with short data sets. At 20 Hz a ten cycle long data set will take 0.5 second. You may need a somewhat longer measurement time or some sophisticated signal processing to get meaningful results in a short time. If your system does not have any significant nonlinearites and a modest set of frequencies is adequate, the multitone system might work.

Lynn

[Henrik_Volkers]

just curious,

the poster say fast and didn't say anything about the quality

The equipment he want to use is low cost (ok, maybe his reference mic isn't)

Shouldn't a single spike out of the speaker (known response assumed ) and captured by the DUT (mic) be enough to calc the the rest?