Can you help me understand why does the HPF change the +\-DC-OffSet ?

[SoundLight]

Hi there, Great Day to everyone


Can you help me understand why/how does the HPF change the +\-DC-OffSet ( to “ 0 “ ) ?


Before HPF:




After HPF:




I am asking because i want to understand -and know- how the audio signal is processed and coded and why and how can HPF work the way it does with the DC

[BertKoor]

Hi there, Great Day to everyone


Can you help me understand why/how does the HPF change the +\-DC-OffSet ( to “ 0 “ ) ?


Before HPF:




After HPF:




I am asking because i want to understand -and know- how the audio signal is processed and coded and why and how can HPF work the way it does with the DC

Links to images do not link to images

[paterpeter]

Consider a sine wave that gets lower and lower. At 0Hz the wave doesn't oscillate anymore but stays at a fixed amplitude with a magnitude depending on the current phase. This is the DC offset. A HPF cuts frequencies below a certain threshold - including 0Hz.

[DJ Warmonger]

DC offset is basically a frequency of 0 - constant signal. So yeah, highpass definitelly removes it.

[JCJR]

As Bert said the question is not clear and so any detailed answer might be answering something you did not intend to ask.

The HP blocks DC. If there is an un-changing DC on the input then the HP output will eventually hit zero and stay zero until the input changes to something else.

A couple of functional examples. If you make a synthetic audio input signal which is zero DC for one second then it jumps to a value of 0.5 for 60 seconds, then jumps back to 0 for 60 seconds. You have 1 sec of all zero, 60 seconds of all 0.5 samples, then 60 seconds of all zero samples again.

Maybe the first order HP filter has a cutoff of 20 Hz or whatever you like. The output when you hit play will be 0 for 1 second, then suddenly jump up to 0.5 and immediately start falling toward zero.

After awhile the 0.5 spike has decayed close enough to zero that it doesn't matter. If you tune the filter lower then it takes longer to decay to zero and if you tune the filter higher it takes shorter to decay to zero.

Then the output would stay zero so long as the input stays unchanged at 0.5. At 61 seconds when the input toggles back to zero, the negative input change causes the HP output to spike down to -0.5 and immediately start ramping up toward zero. After awhile the output reaches real close to zero again and stays there until some other change happens on the input.

On positive sudden dc input changes, a new DC input value bigger than the current input value, makes positive output spikes decaying down to zero. Negative sudden dc input changes, a new DC input value smaller than the current dc input value, makes negative output spikes decaying up to zero.

So a sudden change from DC=1 down to DC=0.5 would give an -0.5 negative output spike even though both old and new input values are positive.

And a sudden change from -1.0 up to -0.5 would give a 0.5 positive input spike even though both old and new input values are negative.

It is only passing changes in the input signal value. It is blocking any steady un-changing inputs. It is passing the difference between the old input value vs the new input value.

Sometimes in analog a simple highpass is discussed synonymous as a differentiator and a simple lowpass is discussed synonymous as an integrator. But there is "nuance" in how those terms are used in that context.

[sault]

A lowpass filter suppresses high frequencies - meaning that it reduces fast changes in the waveform. A highpass filter suppresses low frequencies - meaning that it reduces slow changes in the waveform. The slowest change to a waveform is an offset which doesn't change at all. A highpass filter will remove this, by definition.

[Aleksey Vaneev]

Simply because if you plot a frequency response of high-pass filter, the gain will be zero at DC. Hence it zeroes DC (constant) signal.