Generally I measure sound of machine using B&K microphone. But unfortunately this time I can’t go to field, instead told one of the service engineer to measure the sound of machines using his iphone. When I imported the mp3 file in Audacity the sound amplitude is normalized between -1 to 1. I want to compare the sound amplitude variation at different frequencies for few of the machines.
My question is how the sound amplitude is scaled when mp3 file is imported in audacity? Is there any format in which one can export absolute sound pressure in Pa? Is it possible to scale back the normalized amplitude to absolute sound pressure or atleast the relative sound pressure between different sound files?
There is no direct relationship between “Sound Pressure Level” and the size of the waveform unless you use a calibrated system (such as a calibrated “Sound Level Meter” Sound level meter - Wikipedia). If you have two recordings and one recording is louder (taller waveform) than the other, that gives no indication of which “sound” was louder in real life. The one with the taller waveform may just have been recorded with a higher recording level, or with a more sensitive microphone, or closer to the sound.
You can get a cheap SPL meter for under $50. It wouldn’t be acceptable for testing industrial legal/regulated loudness, but it would be better than using a totally uncalibrated iPhone.
If you use the same setup to record, and the same distance, you can compare the relative levels.
When I imported the mp3 file in Audacity the sound amplitude is normalized between -1 to 1.
There is a little drop-down arrow at the left of the waveform, and if you click that you will see an option for Waveform (dB). But as Steve said, in an uncalibrated set-up there is no known relationship between dBFS in the digital file and dB SPL (acoustic loudness).
However, if one recording is 6dB louder than the other that 6dB difference is valid in both dBFS and dBSPL. Note that zero dBFS is the “digital maximum”, so most dBFS readings are negative (larger negative numbers are “quieter”). Whereas, zero dBSPL is the threshold of hearing and dBSPL readings are positive.
You can test the peak level by running Effect → Amplify and checking the default (then cancel the effect so you don’t actually change anything). But, the correlation between the peak level and loudness is poor. And if the files are clipped, the peak readings are totally useless.
There is an optional [u]stats.ny[/u] plug-in that can give you RMS & A-Weighted RMS values. These will correlate better with relative loudness.
You can get a cheap SPL meter for under $50. It wouldn’t be acceptable for testing industrial legal/regulated loudness, but it would be better than using a totally uncalibrated iPhone.
Sometimes machines are operated at remote locations of the world & you can tell the local service engineer to do best at that moment of time. The data is used for debugging purpose & not comply to any regulatory standards.
Even the relative change in the amplitude between machine to machine tell lot of information considering that data is measured at nearly same location using same iphone for each of the machine.
You can also get a “dB meter app” for an iPhone for about $1. The limitations that Doug outlined applies to these too, but if the same iPhone is used each time it will give a rough idea of relative levels.