Sometimes customers want to do more than simply listen to a WAV recording made by NTi Audio’s XL2.
For customers doing post-processing and analysis on recordings (e.g. MATLAB), the absolute scaling in engineering units (e.g. Pascals) is important. Unfortunately, the generic WAV format does not inherently store data in engineering units, only percentage of full scale, 1.0 to -1.0.
Reading the Full Scale (FS) from the file name
By default, WAV recordings made by the XL2 include the Full Scale (FS) value in the file name (e.g. 2019-10-17_SLM_012_Audio_FS129.9dB(PK)_00.wav)
Reading the Full Scale (FS) embedded in the file
The XL2 stores date and time of the recording within the wav-file (according to EBU TECH 3285 - Broadcast Wave Format BWF). This information is available through professional audio/video tools typically used in broadcast.
Here are some links to various tools for reading the data:
The scaling information is contained in the Description field. Here are examples:
Sound: 0dBFS = 109.7 dBSPL Time Zone: UTC-06:00 (DST)
Vibration: 0dBFS = 2.40 mps2 Time Zone: UTC-09:00 (DST)
For sound data, convert the full-scale value in dB (aka 0dBFS) to a full-scale value in Pascal (Pa) using the formula
Pa = 20E-6*10^(0dBFS/20)
(for example, 109.7 dBSPL converts to 6.11 Pa)
For vibration data, the full-scale value is already in engineering units; 2.40 m/s2 in the example data above.
Simply multiply the full-scale value in PA by the unscaled value in MATLAB. So a data point in MATLAB which was -0.2315 would be -1.41 Pa.
thanks for your helpful post. That's exactly what I was looking for. Maybe you or somebody else can me help with the following question, just for clarification purposes:
In the example at the bottom of your post, is there and decimal point error and shouldn't it be instead either *-0.02314 --> -0.141 Pa* or *-0.2315--> -1.41 Pa* ?
If thats not the case, I would really like to know, what the error in my thinking is. Thank you very much for your help.
Thanks for catching that.
I have corrected my math error and some other items (left out a "not").