Will constantly randomizing the phase of an audio signal, say a speaker in the corner of a square room, reduce standing waves (i.e. room modes) in the room?

For example if you wanted to create a diffuse field (i.e. no standing wave interference patterns) in a small room at low frequencies.


Assuming that you are not changing the phase too much, you are basically turning sinusoids to narrow-band noise. If there is a room frequency response node that strongly attenuates a sinusoid, the total power of narrow-band noise centered at the same frequency will not be attenuated as much. It will survive with a dip in its spectrum. Room modes or frequency response peaks are usually not as severe and narrow as frequency response nodes.

An example speaker and room magnitude frequency response (CC-BY-SA 3.0 by Gregory Maxwell)
An example speaker and room magnitude frequency response (CC-BY-SA 3.0 by Gregory Maxwell)

However, I wonder if your application affords the degradation of audio quality due to the modulation. Also, the room frequency response is not affected so if the test audio is noise, you won't get any sort of improvement.

  • $\begingroup$ Say the application is testing the sound insulation of a wall, where it is desired to have equal energy at all frequencies incident on the wall (to give a truer representation of the wall's performance). So 'audio quality' aside, would the random phase give a more even sound pressure level around the room or would the standing waves and the associated nodes/anti-nodes remain in the same position? $\endgroup$ Apr 20 '16 at 11:55
  • $\begingroup$ If your excitation (speaker output) has a certain power spectrum, and the room has a certain magnitude frequency response, the power spectrum incident on a certain position in the wall will be the product of the two. So unless you change one of them the incident power spectrum remains unchanged. $\endgroup$ Apr 20 '16 at 12:07
  • $\begingroup$ ah ok, so im assuming then that the standing waves are a product of wavelength and room dimensions, therefore phase is irrelevant? I was thinking that the random phase change would randomize the interference patterns and give a more even sound pressure at a single measurement location when averaged over time? Compared to say constant pink noise, which would give a stationary interference pattern and a constant sound pressure at a single measurement location when averaged over time. $\endgroup$ Apr 20 '16 at 12:26
  • $\begingroup$ The reflections will be modulated identically to the direct audio so the interference patterns remain unchanged. One thing you can do is to measure the impulse response of the room–insulation system. Room modes/nodes are caused by wall reflections and these can be greatly attenuated in analysis of the impulse response by multiplying the impulse response with a function that starts with a value 1 and goes to value 0 after the direct peak. Then you would convert the modified impulse response to a modified frequency response by Fourier transform. Be careful how you calculate attenuation. $\endgroup$ Apr 20 '16 at 12:37

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.