What causes an asymmetric waveform in an audio recording?

Status
Not open for further replies.
And why would you need to shift the second harmonic 90 degrees?
The second harmonic is not normally shifted in phase.
What do you mean by "normally"?

With voice and musical instruments, the phase relationship between the fundamental and the harmonics is often such that the combined waveform is asymmetrical. That's why you often had to deal with asymmetrical waveforms in your broadcast career.

The fact that your Adobe software uses a different phase relationship by default is besides the point.

P.S. For anyone not familiar with this, there's some interesting pics of recorded waveforms here.
 
The whole point of this thread is the asymmetry shown on a photo of two voices.
My original point is that our voices are naturally asymmetrical. I'm sure that harmonics play an important role. In my career, I have worked with announcers who had voices which produced absolutely symmetrical waveforms (that is rare).
I have also worked with announcers with voices which were so asymmetrical that I needed to add phase rotators within the microphone preamps to avoid distortion.
I have found that most voices have a 4dB to 8dB asymmetry.
 
As others have found (by experiment) adding even harmonics can result in an asymmetric waveform. It is precisely because of this that voices (or anything else with an asymmetric waveform) are asymmetric. Adding odd harmonics does not do this, whatever their phase - the waveform remains symmetric.

I'm sorry if Frank thinks I am rude, but he directly contradicted my true statement and claimed personal experience authority in so doing. No amount of experience in any profession can rightly contradict mathematical fact.
 
I had always assumed it was a biological reason.

Voice is a combination of sounds created from multiple resonating cavities. It is the interaction of these frequencies that cause the visual asymmetries seen in those examples.

Partial cancellation.

Those waveforms show you the power outline, but the different frequencies are homogenized into one colour so interactions are not visually evident.
 
It's the same thing with instruments, the resonant frequency of the instrument interacts with various notes played causing partial cancellation, mechanically and acoustically.

No need to make it complicated.

I am satisfied with my theory, those who are into this kind of thing may be able to simulate this with software.

Mix simultaneous, amplitude degenerating, random frequency waveforms together. This will be the tricky part as sinusoidal waves won't quite work unless the frequency is continuously varied. (In frequency and amplitude)

A big factor in peoples asymmetrical waveforms would be how much they sing through their nose for example vs a more technical chest and mouth type singer.
 
There's no d.c. offset. It's pure asymmetrical a.c.

Hi,

The only possibility that will reconcile the opinions in this thread
if exhalation at DC is not creating the asymmetry therefore is :

Best way I can describe it is the crest factor of one side of the
waveform is higher than the other side, though in fact both
sides average out in value as they must. Consequently it
makes some sense if that is the case then one side of the
waveform might need treating differently from the other,
or they will in fact react differently to any symmetrical
effect, e.g. compression or limiting, seems fair enough.

This implies the whilst the asymmetry in the OP is real,
the apparent DC offset is an illusion of the envelope.

rgds, sreten.

TBH I don't know enough, but its hard to argue with anyone
who has seen and measured vocal asymmetry. Theory that
fits empirical evidence is much better than trying to refute it.
 
Last edited:
Thank you for your responses!
This all makes sense - harmonics can be introduced when air passes through various cavities. I get that the human voice does not necessarily have to produce a symmetric waveform. Even looking closely at the rest of the waveforms it's clear that they are not symmetric on a finer scale.
Since the distortion in the beginning is so obvious I was hoping that somebody could recognize what might have caused it. I thought it might be a recording equipment malfunction but it seems like it's just the way it was pronounced.
 
Well, the total energy over time will be equal.

An impact, be it an initial higher pressure, or initial low pressure, will be asymmetrical over the first few cycles. Again, very quickly the " energy under the curve" has to be the same or else there would be a static rise in air pressure. Go look at a record scratch.
 
Status
Not open for further replies.