significance of a waveform above the zero line

[paulpoitier]

Hi all,

I'm trying to understand how to understand sound wave graphs, and there's one thing that I can't find an answer to anywhere online.

I know that the frequency (which we hear as pitch) is represented by how condensed the waves are horizontally and that the amplitude of a wave (which we hear as loudness) is represented by the distance between the peak and the trough of a wave, vertically.

I also understand that the amplitude represents how condensed the air particles - higher = more condensed physically, lower = more spaced out.

However, I cannot find an answer to what it means when a wave is entirely above or below the "zero line". For example, a waveform which has an amplitude of between +3dB and +2db. What is the difference between this and a wave with an amplitude of -3dB to -2db? To my mind, both have an amplitude of 1dB, so would be the same loudness.

I hope this makes sense, and thanks in advance.

P

[ecuk]

This is known as a DC offset (or a number of other similar terms): https://en.wikipedia.org/wiki/DC_bias.

It is generally bad in audio, but it can be removed by subtracting (or adding) the DC offset from every sample. After doing so the ‘new’ DC offset will be zero.

Edit: That said, you are correct that a DC offset is usually considered inaudible. It only becomes a problem when there is subsequent processing of the signal (which in a DAW there often is).

Edit 2: Here is another good description of what causes it, potential problems, etc.: https://manual.audacityteam.org/man/dc_offset.html

[paulpoitier]

This is known as a DC offset (or a number of other similar terms): https://en.wikipedia.org/wiki/DC_bias.

It is generally bad in audio, but it can be removed by subtracting (or adding) the DC offset from every sample. After doing so the ‘new’ DC offset will be zero.

Edit: That said, you are correct that a DC offset is usually considered inaudible. It only becomes a problem when there is subsequent processing of the signal (which in a DAW there often is).

Edit 2: Here is another good description of what causes it, potential problems, etc.: https://manual.audacityteam.org/man/dc_offset.html

that’s super interesting, thank you for taking the time to explain. it seems that when i look at every musical waveform close-up, they all seem to have many points like this - here is an image of a snare sound, for example - https://www.researchgate.net/figure/323_fig2_329047208 - and you can see that the waveform is not centered around the zero line, and some parts are entirely above or below. so that would mean that those parts are inaudible?

[ecuk]

Hmm, no, not exactly. Every waveform will have some parts above the average and some below the average. This is, after all, the nature of averages. And the DC offset is, in essence, just the average amplitude of a waveform.

So it isn’t that the parts above and below the average (i.e., above and below the DC offset) are inaudible. It is that the average offset makes no difference to what you hear. You could shift a waveform up or down without affecting the perceived sound. (Provided you do not clip the signal when you shift it.)

The problems arise because subsequent processing generally assumes that the DC offset is roughly zero. If it isn’t you are might end up with unwanted clipping, distortion, etc. By the way, you can remove DC offset in Live by enabling the DC button in the lower right of a Utility device.

Another way I like to think about it is this: your ears respond only to the relative changes – the differences – in the waveform from one moment to another, not to the absolute amplitude at any given point in time. There is no such thing as a sound at a single point in time.

[paulpoitier]

Hmm, no, not exactly. Every waveform will have some parts above the average and some below the average. This is, after all, the nature of averages. And the DC offset is, in essence, just the average amplitude of a waveform.

So it isn’t that the parts above and below the average (i.e., above and below the DC offset) are inaudible. It is that the average offset makes no difference to what you hear. You could shift a waveform up or down without affecting the perceived sound. (Provided you do not clip the signal when you shift it.)

The problems arise because subsequent processing generally assumes that the DC offset is roughly zero. If it isn’t you are might end up with unwanted clipping, distortion, etc. By the way, you can remove DC offset in Live by enabling the DC button in the lower right of a Utility device.

Another way I like to think about it is this: your ears respond only to the relative changes – the differences – in the waveform from one moment to another, not to the absolute amplitude at any given point in time. There is no such thing as a sound at a single point in time.

that last sentence of your above response is really fascinating, thank you, i've been watching some old bbc videos about synthesis and they never pointed that out. i guess the heart of my question still remains - what is the difference between a waveform which is offset above the zero line and one below it, if they are both of the same amplitude and frequency? in other words, how are particles of air affected differently by a positive or negative DC offset? and if they are not affected differently, what does the offset represent in the real world? thanks a lot for your time, by the way

[fishmonkey]

Hmm, no, not exactly. Every waveform will have some parts above the average and some below the average. This is, after all, the nature of averages. And the DC offset is, in essence, just the average amplitude of a waveform.

So it isn’t that the parts above and below the average (i.e., above and below the DC offset) are inaudible. It is that the average offset makes no difference to what you hear. You could shift a waveform up or down without affecting the perceived sound. (Provided you do not clip the signal when you shift it.)

The problems arise because subsequent processing generally assumes that the DC offset is roughly zero. If it isn’t you are might end up with unwanted clipping, distortion, etc. By the way, you can remove DC offset in Live by enabling the DC button in the lower right of a Utility device.

Another way I like to think about it is this: your ears respond only to the relative changes – the differences – in the waveform from one moment to another, not to the absolute amplitude at any given point in time. There is no such thing as a sound at a single point in time.

that last sentence of your above response is really fascinating, thank you, i've been watching some old bbc videos about synthesis and they never pointed that out. i guess the heart of my question still remains - what is the difference between a waveform which is offset above the zero line and one below it, if they are both of the same amplitude and frequency? in other words, how are particles of air affected differently by a positive or negative DC offset? and if they are not affected differently, what does the offset represent in the real world? thanks a lot for your time, by the way

it depends. in most cases a signal with a modest DC offset may cause no noticeable audible difference. think about how a sound wave is generated by the most common types of audio transducers, i.e. speakers. speaker cones vibrate back and forth to create the pressure waves in the air. with a DC offset, the "zero" point of the speaker cone vibration will be offset from the zero resting point of the speaker cone. this probably won't cause any issues unless the DC offset is so large that the speaker can no longer faithfully track the waveform...

[ecuk]

What fishmonkey said.

I would add to that still another way of thinking about it. The ‘absolute’ air pressure – think weather – does not appreciably affect the sounds you hear. Sitting inside your studio, a given waveform will sound essentially the same regardless of whether there is, outside the window, a lovely day with clear skies with a fairly high pressure of 1020 mbar or a rainy and overcast low pressure system at 980 mbar. Either way what matters are the much smaller differences – the changes, modulations, fluctuations, oscillations, call them what you will – that happen from moment to moment between the sound source and your ear (or microphone, etc.). The underlying air pressure could thus be thought of as a DC offset, in this case the baseline air pressure at the eardrum.

Or perhaps another way to think about it. We are used to thinking of silence as a flat line at zero, right? But the ‘zero’ part of this is entirely arbitrary. If you shift the level of this flat line either higher or lower, you will still hear utter silence. The flat line is what gives you silence, not what level – speaker cone displacement, air pressure, voltage – that flat line is at.

Now add some bumps and wiggles to that flat line. The same thing holds true: the waveform, in this case not representing silence but something you can hear and maybe even enjoy, will sound the same regardless of the average level – speaker cone displacement, air pressure, voltage, whatever – that those bumps and wiggles are at.

[paulpoitier]

I understand, and I feel like my viewpoint of life and my physical brain size have enlarged. thanks both for your help - the real life analogies were especially vivid

[ecuk]

You’re very welcome, and thank you for asking the question.

It’s fun to have an excuse now and then to take a moment and think about how to say things that I too often take for granted

[fishmonkey]

while we are geeking out, from our sensory faculties upwards we have evolved to notice difference, change, contrast. we tend to habituate (become desensitised) to things that we experience continuously. sharp unexpected discontinuities are especially noticeable to us.

in the music production context, this is why relatively minor glitches in audio playback bug us so much and music production forums always have people complaining about their playback system not being 100% rock solid. it's also why a classic mixing technique is to bring new elements into a mix a little too loud so they poke out, and then fade them back down into the mix afterwards. or why drops are a thing in EDM...