I would like to get the ambient noise levels of a habitat that I have been monitoring for some time (monitoring birds, 0-16 kHz). For this, I want to calibrate my recorders and get their sensitivity and frequency response and then use my calibrations on the recordings to get to the noise levels.

However, I still didn't manage to come up with a workflow that would enable me to reproduce the sensitivity measurements very well. Does someone have an idea/solution to help me with :

  • making sure that the recorders are all exactly at the same spot every time and facing the same direction (until now, attaching them to a tripod always gave rise to some variation in the direction the microphone was facing, even with a lot of care)
  • reduce as much as possible the reflections from tripods around the microphone: what would be an appropriate distance between the loudspeaker and the microphone for audible sound?
  • analysis: what would be an acceptable measurement error? let's say that I have a 3dB difference between two sensitivity measurements of the same microphone two weeks apart, is it fine? or should I aim for smaller differences?

Finally, is my workflow OK? or are long-term recordings too variable (due to e.g. temperature variation, battery level, etc.) to extract anything reliable?

Any experience/tips and tricks to do good and reliable microphone calibration are welcome!

  • $\begingroup$ "what would be an appropriate distance between the loudspeaker and the microphone for audible sound": do you mean a sound source instead of a loudspeaker? I understand from your first paragraph that you are recording natural sounds in the field, aren't you? $\endgroup$
    – Noil
    Sep 28 at 6:53
  • $\begingroup$ I am recording natural sound in the field but I came back to the lab for the calibration $\endgroup$
    – lframond
    Oct 10 at 7:55

1 Answer 1


In a free field, the attenuation is 20 log10(d) where d is the distance between the source and the receiver. This means that relative positioning errors will have more impact on the calibration as the recorder to be calibrated gets closer to the source. You can evaluate the uncertainty u on the measurement of distance and try to select d so that 20 log10((d+u)/(d-u)) is low enough.

But there are certainly other causes for uncertainty in your calibration setup and procedure: is your sound source small enough to avoid directivity effects? What is the signal to noise ratio during the calibration? How do you control the output of your sound source? Do you bring your recorders to the lab to calibrate them or do you calibrate them in the field?

A 3 dB difference between two successive calibrations seems quite large. In acoustic measurements standards, a common rule is that a measurement should be discarded when the drift between the calibration done at the beginning and the one done at the end is above 0.5 dB. But one should bear in mind that automatic recorders are not designed to measure sound pressure. So this requirement is probably too strict in your case.

  • $\begingroup$ Thanks for the answer, I am doing calibrations in the lab, in an echo-attenuated room. At the moment I use a sound level meter and a recorder and a calibrating microphone at the same time, 1m in front of a loudspeaker. But somehow I have a lot of variability between two measurements and I can't get everything to stay reproducible. Maybe my loudspeaker is too loud and I still pick up some reflexion from the ground? What would be a "good" level for a calibration? I get sometimes measurements that are as good as 1 dB apart and sometimes as bad as +/-5 dB, and I can't understand why as of now. $\endgroup$
    – lframond
    Oct 10 at 8:02
  • $\begingroup$ You will get a reflection from the ground whatever the playback level. Let's forget the device you want to calibrate for a moment. Did you check the stability of your sound source? What happens when you place the sound level meter in front of the source and do repeated acquisitions without moving the sound level meter (SLM)? If your levels do not change, what happens when you (1) place your SLM, (2) measure the sound level, (3) remove the SLM several times in a row? What signal do you use for calibration at what sound level typically? Do you calibrate your SLM with a 1kHz calibrator? $\endgroup$ Oct 11 at 10:46
  • $\begingroup$ The sound source is very stable when I don't move the SPL meter, (within 0.2 dB), and when I move the SPL meter it is not too bad either (within 1 dB). I had used a source level of 48 dB @ 1 m @ 1 kHz in my echo free room which had an ambient level of 32 dB, but the signal at different frequencies ranged between 40 and 53 dB due to the frequency response of the loudspeaker. I did not calibrate the SPL meter, next time I'll do it $\endgroup$
    – lframond
    Oct 17 at 7:08

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.