I am analyzing the spectrum of low-level signals and try to calibrate the system, to get the correct dBSPL levels in the results.
The only sound pressure level calibrator I have at hand generates a reference signal at 94 dBSPL at 1kHz. I cannot zero the recording level at this level because the measured signal is too low to get useful recordings.
I thought I could just calculate the difference between two reference signals:
- I use two identical calibrated microphones, with exactly the same characteristics.
- I connect both microphones to the same recording device as A and B.
- I first set the recording level for A and B to the same level and check the recording level with the 94 dBSPL reference signal. Both microphones return the exact same signal.
- At this point:
- in my recordings of the reference signal -6 dB equals a sound pressure level of 94dBSPL
- therefore the 0 dB in my recordings equals 100dBSPL.
- Next, I change the recording level for microphone A to the required setting to record the low-level signal.
- With both microphones connected, A on the higher target level, B on the calibration level, I record a 1 kHz signal, from a speaker at the exact same distance to both microphones. The sound pressure of this signal is unknown at this point. The signal sound pressure is adapted to get a good recording at the target recording level for microphone A.
- Microphone A at target recording level, recorded the signal at -3 dB and microphone B at the calibrated recording level recorded the same signal at -25 dB (rounded).
- As I assume the 0 dB point at the calibrated level is 100 dBSPL, the recorded signal had a sound pressure level of 75 dBSPL at the tip of the microphone.
- As this 75 dBSPL signal was recorded with -3 dB at the target level, I assume the 0 dB point of the target recording level is 78 dBSPL.
Is this a valid way to get the approximate sound pressure of the zero point for the recording level at 1kHz?
My calibration does not haveto be super precise, it is enough to be in the range of +/- 5dB to allow a later dB(A) conversion and see which parts of the signal are audible for humans.