Signal going below the zero line in audacity

[Rockin Mike]

So, I'm using audacity to record from the board and I have always wondered why half of the waveform is above zero and why half is below. When it's below zero does that mean there's negative sound? Is there such a thing as negative sound?

I noticed today that the above and below parts are fairly symmetrical until I cranked the gain on the channel which drove the board into the red and produced a heavily distorted signal which looked like this in audacity:

What is going on with the signal here?

I'm not looking for "don't crank the gain"... that was an experiment and the power amp was turned down to protect the innocent.

 

[callofcthulhu]

That image would suggest there's a fault somewhere in your signal chain/interface. Generally speaking there should be about as much signal below the 0 crossing as there is above. This DC offset represents the halfway point of a single cycle of sound, which is represented as a sine wave (which you may remember from your trigonometry classes looks a bit like a sideways S - think of the 0 point as the line you cross if you were to draw a perfectly symmetrical S).

In practical terms, just listening to a file with a wonky offset the naked ear probably won't hear any difference. However, all of the DSP algorithms in any DAW you're using are going to run based on the assumption that you have a DC bias centered at 0, and will behave strangely if you don't. Also, if the left channel is offset differently than the right channel of a standard two channel stereo file - which your example looks like it may be - I'd imagine that sound file sounds like it's phasing in some weird way, maybe too subtly for you to notice.

Your image suggests the left and right channels had their sample rates misaligned. Either they're sampling at different rates or didn't have their start points sync'd (possible if each signal is coming from a different interface or your interface is broken), or you ran out of memory and Audacity undersampled the right channel to prevent a crash (possible if you need more RAM). I'd check that all your input devices are sync'd to the same clock, and that Audacity is too. In the mean time, under the effects tab there is a process called "Remove DC offset" which will realign your 0 crossings.

 

[Rockin Mike]

Yeah that's an extreme example, it's usually symmetrical around the zero line or pretty close.

So, when the signal is going back and forth around this zero point, what is being graphed? Is this voltage? Sound pressure level? Speaker excursion (well obviously not that)? There are no labels (that I can see) on the audacity scope.

 

[khutch]

So, when the signal is going back and forth around this zero point, what is being graphed? Is this voltage? Sound pressure level? Speaker excursion (well obviously not that)? There are no labels (that I can see) on the audacity scope.

Audacity works on sound files that have been converted to numbers so the graph you see on your screen is just numbers. There are no units. Originally the numbers may have been sound pressure levels or voltages depending on whether the recording was made with microphones or a direct electrical connection. The pressure level at the microphone varies cyclically both higher and lower than atmospheric pressure (which we arbitrarily call zero) when sound waves pass the microphone. The voltage coming out of your bass varies cyclically above and below zero volts when you play it. So the numbers on the Audacity graph also vary cyclically around zero. The reason your graph shows the signal entirely above or below zero at times is that there is low frequency noise on the signal that is larger in amplitude than the sound that you can hear. I have no idea where it is coming from.

 

[callofcthulhu]

All of the above and none of the above, really. The SPLs are translated into voltages by your microphone/pickup whatever, which are then converted into binary numbers by your A/D converter, which are then recorded by Audacity.

Again, remember that sound is a wave. There are no peaks without valleys.

For all intents and purposes, think of 1 and -1 as the points of maximum signal before clipping occurs. In the case of Audacity it's the largest/smallest binary number a single sample can be represented by, which will be smaller or larger depending on your bitdepth.

Imagine a sound wave hitting your microphone. The diaphragm will get pushed in according to how powerful that wave is, and that depth will be translated into some positive bit string. As the wave propagates, there is a vacuum of some magnitude that will then suck the diaphragm the opposite direction. This is your negative value.

The rate at which the diaphragm is pushed and sucked is the frequency of your sound, the amount it's pushed and sucked is your amplitude.