Frequency isolation

Hi there

I’ve been taking recordings of a ringing outside my house which is need to find the volume of. However, there’s lots of background noise in the recordings which means I can’t take the volume reading accurately.

I know the specific pitch of the ringing (1162 Hz) , so is there any way to isolate that in the track and remove all other noise?

Thank you


There is no way to measure volume (or loudness), as your setup isn’t calibrated. And since the signal you want to measure drowns in the background noise, there isn’t even a way to measure it uncalibrated.

The dB level scale in Audacity is relative. For a loudness measurement you need to set it to an absolute reference, as provided by a calibrated measurement microphone and these tend to be expensive. One of the least expensive ones is the UMIK-1. See:

I’ve never tried it and I am a bit skeptical about the calibration claims…

which means I can’t take the volume reading accurately.

I assume you already have an SPL meter?

…so is there any way to isolate that in the track and remove all other noise?

A bandpass filter.

Apparently Audacity doesn’t have a bandpass filter, but you can make one by combining high-pass and low-pass filters. For example, apply a high-pass filter to kill (actually reduce) everything below 1162 Hz and then apply a low-lass filter to kill everything above 1162 Hz.*

Here’s something you can try -
1. Make a 1162 Hz sine wave with Audacity’s Generate feature.

2. Set-up a speaker, your SPL meter, and your computer microphone outside where you want to measure the offending noise.

3. Play the test/calibration tone loud enough to drown-out the offending noise and the other noises so you’re sure you’re recording & measuring the loudness of the test/calibration tone. Record the sound and note the SPL reading on your SPL meter.

4. Without moving anything or changing any recording settings, turn off the test/calibration tone and record the offending noise (along with the other outdoor noises). There is no need to measure the SPL this time.

5. Open the offending noise recording in Audacity and apply the bandpass filter. Look at the waveform and check the dB level** in Audacity. You can check the peak or try to mentally average, or ignore outlier peaks, or whatever seems to make sense…

6. Open the recorded test/calibration tone and apply the same bandpass fitering. That may not be necessary, but you should process both files the same. Note the dB** reading in Audacity.

You now have 3 decibel readings. You have the SPL loudness of your test tone, the digital dB level of your test tone, and the digital dB level of the (filtered) offending noise. You can now correlate/calibrate the digital dB reading with the dB SPL reading.

For example, let’s say the recorded test-tone is -10dB and the test-tone SPL reading is 80dB. If another recording is -20dB in Audacity, that’s 10dB less than the calibration, and the SPL loudness is also 10dB less for a loudness of 70dB SPL.

8. Subtract the recorded offending-noise dB level from the recorded test-tone dB level. Subtract*** that same dB difference from your test-tone SPL level to find the SPL level of the noise. (Try not to get confused when subtracting negative dB levels :wink: )

NOTE - Loudness measurement is tricky stuff. None of the above would be good enough for legal or regulatory purposes.


  • Note that filters are defined by their -3dB point. So, applying a 1162Hz high-pass filter will reduce 1162Hz by 3dB and a low-pass filter will reduce it by another 3dB. (You might be able to live with that, but it doesn’t make an ideal bandpass filter.) So, it’sbest to leave a “gap” by decreasing the high-pass frequency and increasing the low-pass frequency.

** You may have to use the Amplify effect to “see” the dB level accurately. Just keep track of the change so you know the real-original level.

*** Of course, if the offending-tone was louder than thbe offending tone you’d add to the calibration SPL level.

A possible command for the Nyquist prompt:

(bandpass2 *track* 1162 80)

Testing with a chirp tone: the peak seems to be preserved.