TOPIC:

Hello,

I want to do long-term measurements with the UMIK-1. When I use the logger in REW, I do not have information about the frequencies, which I want to inspect, too.

So my alternative is to use Audacity to record the raw signal, compute one-third-octave bands, weigh them with the A filter (via Matlab scripts) and combine them for an SPL reading. But how do I incorporate the calibration file? What does the "Sens Factor" mean? How do I adjust for it? Which value do I choose for the 1/3-octave bands in the cal file, only the approximate center frequency or do I average the adjustments for the width of the band?

Any advice will be greatly appreciated.

@ Honas,

Sounds like a unique project ' . A bit outside our tech support structure here so just some pointers to get your started, you'll have to do more homework/research on your side to figure out the rest.
- Calibration file: You'd have to go to frequency domain I guess (FFT on your recording). It's similar to the path you're doing A weighting. Just with a different response.
- Sens factor is only for SPL value. Not sure how easy you'll be able to reverse engineer that. That sens factor is basically the dBfs equivalent value that the driver is seeing when a 94dBSPL is played into the mic.

DevTeam

Hello DevTeam,

first of all, thank you for your reply.

- Transforming into the frequency domain is no problem. But I use linear steps, e.g. 2^16 equally-spaced frequency values for my x axis (from 0 to 20 kHz). The provided values in the calibration file do not have the same spacing (they are logarithmic). So I need to interpolate. I can then use an inverse FFT to get my adjusted Signal. EDIT: But those calibration values are in dBSPL I assume, which leaves me with the question on the very bottom.

-Sens Factor: I've read in another post that this value affects the value "HKEY_CURRENT_USER\Software\JavaSoft\Prefs\room eq wizard\last/Spl/Offset" in the registry. The sens factor is subtracted of 120 dB, e.g. my sens factor of -.673 dB yields a total of 120.673 dB in the registry. Is that value the full-scale (dBFS = 0)?
I am not sure how "-.673 dB" corresponds to the dBFS equivalent value @ 94 dBSPL. I think when you use a calibrator with a 1 kHz sine tone @ 94 dBSPL, then you need add the sens factor to your dBSPL to get it calibrated to 94 dBSPL (else you would get 94.673 dBSPL).

-Do I need to set the Windows mic recording level to 100% or 50%? I am using Windows 7, SP 1 (32 bit).

-How do I compute my own dBSPL? The values of the FFT should also be in dBSPL, so I can adjust them. I think I can relate my digital signal to the full-scale value, somehow.

@ Honas,

Reaching out the limits of our support here I"m afraid (e.g. how REW registry key are affected.. not sure if you're aware but we're not at all related to REW..
- Calibration settings are in dB. That's a universal scale. You don't need SPL reference until the end I guess when you do calibration.
- Registry question: Completely outside our support I'm afraid..
- Windows volume: you do need to figure this out (e.g. REW locking control). Using a 3rd party SPL meter to confirm the data would be the way we'd do I guess.

Good luck for your project.

DevTeam

Hello DevTeam,

I understand that you are not directly affiliated with REW.
Your last comment about the frequencies in dB (and not dBSPL) was very helpful. I used 20*log_10(x) to transform my frequencies (via a normalized FFT) into decibels and adjusted my spectrum.

Final Questions:
- The calibration settings need to be subtracted, correct?
My reason: The lowest and highest frequencies of the calibration are attenuated, just like you would expect in a microphone. And you want to equalise the frequency response.

- How do I relate my digital amplitude from -1.0 to 1.0 to sound pressure level (SPL)?
For SPL you need the RMS value of the signal, which you approximate at a timepoint by time-weighting ('Slow' or 'Fast'). So now my value is between 0 and 1.0. A sine wave with peaks at -1.0 and 1.0 is defined as 0 dBFS (with an RMS of ca. 0.7072). I suspect that my sens factor of -.673 dB delivers a full-scale value of 120.673 dBSPL (dBFS = 0). So when my RMS is 0.7072 I know I had an of input 120.673 dBSPL, correct?

-In the technical specifications, what does "output noise level at max IGPA : -74 dBFS" mean? Is it the dynamic range?