I’ve really been reading quite a bit and I do hope I haven’t missed just one page that would throw some light on all of this. Can someone help me understand what I’m looking at here or point me to a useful reference?
Many charts tell us 0 dB is the threshold of hearing and 130 dB is the threshold of pain (and so on). But zero dB is the max for recording without clipping in Audacity; and we are always measuring the distance below that ceiling. What is the relationship between these two scales? It seems they must overlap. Can one translate between them as between Fahrenheit and Celsius?
When looking at the linear scale on a track in Audacity, what is the unit of measure of amplitude? The wave is at 0.1 or 0.5 or 1.0 of what exactly? Is this related to RMS? a measure of distance from zero? In what units is distance measured? Or is there some relationship between the linear scale and decibels?
There are many different “dB” measurements. The term “dB” on its own is actually pretty meaningless until the kind of “dB” is either specified, known or implied. Try reading this: http://en.wikipedia.org/wiki/Decibel
Zero dBFS is the maximum level, above which clipping (= audible distortion) will occur. Audacity’s waveform scale value of +1 equates to 0dBFS.
Many charts tell us 0 dB is the threshold of hearing and 130 dB is the threshold of pain
That would be dBSPL. Sound Pressure Level. And this one has variations: flat, tailored to your ear, etc.
zero dB is the max for recording without clipping in Audacity…
That one is dBFS. Full Scale, and it’s not just Audacity. Everybody does it that way. 0dBFS is where the digital measuring system runs out of numbers and can’t get any louder (in general).
The Audacity scale is in percent because it happens to be convenient to edit like that. 1.0 = 100%.
They’re all peak or peak-to-peak. We haven’t measured RMS since the last ANSI C16.5 VU meter went out of service, and even that didn’t conform to RMS exactly because it had a ballistic constant.
Koz
Missed one. For all those people who want to measure room noise with Audacity, you can’t. There’s no direct translation between dBFS and dBSPL. You need to break down and buy a sound pressure level meter.
Thank you so very much. Just knowing a couple of extra terms can help so much.
Koz, I accept that I cannot measure room noise with Audacity. But it seems inevitable that 0 dBFS must lie somewhere on the dBSPL scale, somewhere between hearing threshold and a jet engine roar. No? That is a different question. Even an approx. or rule of thumb would be welcome. But the question is more about intellectual curiosity and feeling good at this point. I see why I have no practical use for this info.
And PGA, I had looked at wikipedia. (Do a google search and what comes up first, after all.) Like many wikipedia articles, that one starts out with a lot of knowledge (and recollection about math) assumed and perhaps tries to be too comprehensive. I am always grateful; they permit me to formulate questions. dbSPL is rather easier to find and digest. But entries under audio electronics, don’t come close to the kind of clarification you two supplied.
Yes, we can (Obama says), but to compute the dbSPL from the dBFS you need the exact electrical amplitude, frequency, and phase response of your microphone, your analog preamplifier (maybe part of your soundcard), and the A/D converter of your soundcard. With exact I mean not just simply looking-up the diagrams in the manufacturer’s data sheets because they only contain the statistical average of the entire production series, instead you must take measurement equipment (many times more expensive than a simple SPL-meter) and measure these data for your specific equipment. Only then you can compute the dbSPL from the dBFS backwards through the entire transmission line.
You see, it’s possible, but buying an SPL-meter is the better idea.
There is no rule of thumb because the electrical data of microphones and preamplifiers can vary wildly in a range from 1 to several million. Tolerances of several hundred percent are very common in electronics. That’s why measurement equipment is expensive. Producing 1 percent tolerance needs several million times more effort than building a simple soundcard.