Hey all,
I have a project which I have been having a bit of trouble implementing. I'm performing several measurements at 48kHz/24Bit of some sound events. My requirement are as follows for sound pressure spectra analysis.
- 500 - 20 kHz frequency bandwidth with a 25 Hz frequency resolution.
- Peak Hold averaging with 50% overlap processing and Hanning windowing applied to the time records.
- Anti-aliasing filtering.
- A - weighting.
I have the sound and vibration toolkit and have implemented as follows:
1. Read wav
2. Convert from Volts to dB via SVL Scale Volts to EU.vi. (I used a calibrated signal of 94dB at 1kHz to find my mic sensitivity).
3. A-weighting using SVL A,B,C Weighting Filter.vi
4. Peform analysis using SVL Zoom Power Spectrum.vi.
Few Issues:
- I dont know how to implement the Anti-aliasing filter.
- Is the method I have used ok with measuring sound spectra? Or should I perform the FFT then convert to dB?
- I have tried to implement the same analysis using Matlab and although I get the same FFT profile, my dB levels are different where the 16kHz noise is almost the same level as the 8kHz noise? Not sure what is happening there, but I want to concentrate on using LV for now.
I dont have any guidance available or experience with this sort of analysis. I have trialed my program out on a sample wav file which has two separate distinct noises events, one at 8kHz and another at 16kHz. My FFT is showing that the 16kHz is a much higher sound level (95dB) than the 8kHz (80dB) but when I listen to the wav via a pair of studio monitor speakers, the 8kHz is much more audible. Wouldn't the implementation of A-weighting filter tend to make the 8kHz noise more prevalent in the FFT?
Any suggestions on this topic are appreciated.
Thanks!
