I am learning more about DSP specifically for audio measurement purposes.

I can an app, where I can create a log sine sweep from start frequency to end frequency over a certain length. I can perform an FFT on this signal, and observe it is decreasing in magnitude logarithmically, and I can apply an inverse of the sweep to the FFT to correct this, and view the 'flat' frequency response.

If I view the impulse response (both X and Y are the same stimulus signal for now) I can see a perfect vertical line as this is an electronic measurement.

I then make two measurements, one from 20Hz - 150Hz (to simulate a Subwoofer) and 150Hz - 20000Hz for the tops. I can view the FFTs, of these two measurements, however the power of the low measurement is a lot higher than the power of the tops because there is more energy in the stimulus as it is over the same length.

When making these LSS measurements, I am using the inverse of the sweep to correct the logarithmic frequency magnitude, however should I be applying a time gate for these frequencies? ie, I know when a specific frequency is generated and using the impulse I can calculate the lag (none in this case as its the same signal) between measurement and stimulus, so using both of these I can filter anything arriving outside of this?

Or should I be applying a multi time window which very broadly divides the spectrum into several smaller FFTs at different window lengths? (If this is the case, how do I apply this so there is smooth overlap and retain phase over the entire spectrum?

Any pointers would be great, as I can find similar questions asked, but no real answer for how I should continue.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.