vojko v

I am a rookie in these matters and I'm trying to wrap my head around the all-pass filter. I have a few questions:

I ran into a tutorial for Schroeder's reverberator and this is the diagram that was used:

schro.png

If I understood correctly the all-pass filter only has effect on the phase of the sound, and to "hear" the effect you have to mix it with the original clean signal. So what purpose do those three all-pass filters connected in series even have in this diagram without the clean signal? The diagram is from here: https://ccrma.stanford.edu/~jos/pasp/Schroeder_Reverberators.html

2. On repeated watching of the tutorial I was even more puzzled. I don't know how to read these diagrams but the way the video creator applied the logic from the picture into gen~ doesn't make sense to me at all, he is feeding signal with gain b0 back to the delay, but from the diagram delay line should be fed back to itself via the -Am coefficient (+ the input signal), and this sum is then going through the b0 and summed with the output of the delay line. Also the way he used the factors from the diagrams seems just wrong to me, can someone correct me or confirm this is wrong?

https://www.youtube.com/watch?v=F_FhvxT5mLA&t=883s&ab_channel=tune4media

3. I tried experimenting with [allpass~] in Max and was confused even further. I made a patch that has a a [saw~] going through an [allpass~], and yes the [allpass~] doesn't change the frequency spectrum of the [saw~] BUT only for higher frequencies: if I change the frequency of the [saw~] to 20Hz the [allpass~] drastically changes the sound, even without a clean signal being mixed in.

message

spectroscope~

live.dial

m4l.cross1~

flonum

scope~

saw~

ezdac~

live.gain~

allpass~

I have a feeling maybe I am fundamentally misunderstanding what does it mean to change a phase of a signal. I figured it just means to displace it a few milliseconds later.

1.
I ran into a tutorial for Schroeder's reverberator and this is the diagram that was used:


2. On repeated watching of the tutorial I was even more puzzled. I don't know how to read these diagrams but the way the video creator applied the logic from the picture into gen~ doesn't make sense to me at all, he is  feeding signal with gain b0 back to the delay, but from the diagram delay line should be fed back to itself via the -Am coefficient (+ the input signal), and this sum is then going through the b0 and summed with the output of the delay line. Also the way he used the factors from the diagrams seems just wrong to me, can someone correct me or confirm this is wrong?

I have a feeling maybe I am fundamentally misunderstanding what does it mean to change a phase of a signal. I figured it just means to displace it a few milliseconds later.


I am a rookie in these matters and I'm trying to wrap my head around the all-pass filter. I have a few questions:
1.
I ran into a tutorial f

few-questions-about-the-allpass-filter

hehe. yeah, phase only. but that also delays the signal. so what you see here would refer to as the "first reflections" of the reverb algo.

the guy in the video uses a delay buffer to recreate an allpass. this is of course not very useful to understand why and how you typically use allpasses.

the allpass is kind of abused in a schröder reverb.

ask yourself how often you have 20 Hz material in your music, and then change your test setup accordingly.

also, i would not use delay times smaller than 5ms (a room reflection in a room of smaller than 2,5 * 2.5 meters?) for such a test setup. use something more known to your ears (and because phase matters as soon as you sum with the original.)

1. 
hehe. yeah, phase only. but that also delays the signal. so what you see here would refer to as the "first reflections" of the reverb algo.

2. 
the guy in the video uses a delay buffer to recreate an allpass. this is of course not very useful to understand why and how you typically use allpasses.

3.
the allpass is kind of abused in a schröder reverb.
ask yourself how often you have 20 Hz material in your music, and then change your test setup accordingly.
also, i would not use delay times smaller than 5ms (a room reflection in a room of smaller than 2,5 * 2.5 meters?) for such a test setup. use something more known to your ears (and because phase matters as soon as you sum with the original.)


thanks man. is there some almighty DSP book that covers this type of stuff?

few questions about the allpass filter