The feedback type switch in our MagicDelay is a new concept and probably not that straight-forward to grasp. While you can (and you should) adjust this by ear, I’m going to shed some light on this answering the following questions:

  • What is dynamic feedback?
  • Why do I need this?

A picture is worth a thousand words

Let’s have a look at the conventional way of controlling feedback in percent first. In the graphs below the x-axis represents time and the y-axis represents level in dB:

Same delay time, various feedback in percent

Same delay time, various feedback in percent

(a) Input: To keep things visually simple we use a single sample impulse as input. Imagine a really short click sound.

(b) Input + Delay with 0% Feedback: As expected we get a delayed impulse and nothing else because we are not applying any feedback at all.

(c) Input + Delay with 50% Feedback: Each time we feed back the signal we reduce it by 50%, which translates to about -6dB per iteration.

(d) Input + Delay with 75% Feedback: Each time we feed back the signal we reduce it by 25%, which translates to about -2.5dB per iteration.

(e) Input + Delay with 90% Feedback: Each time we feed back the signal we reduce it by 10%, which translates to about -1dB per iteration.

So far this shouldn’t be that surprising. Notice though, that we are always using the same delay time while changing the feedback in percent. Let’s see what happens if we keep the feedback constant while changing the delay time:

Same feedback in percent, various delay times

Same feedback in percent, various delay times

(a) Input: To keep things visually simple we use a single sample impulse as input. Imagine a really short click sound.

(b) Input + Short Delay with 50% Feedback: This is the same delay time and feedback as (c) above.

(c) Input + Medium Delay with 50% Feedback: Using a medium delay time with the same feedback amount the tail gets longer.

(d) Input + Long Delay with 50% Feedback: Using a long delay time with the same feedback amount the tail gets even longer.

(e) Input + Very Long Delay with 50% Feedback: Using a very long delay time with the same feedback amount the tail gets very long exceeding our graph even.

We always apply 50% feedback, so we always get about -6dB per iteration. Since each iteration is spaced apart depending on the delay time, the tail will grow with the delay time.

Within a conventional delay effect this might not be an issue, but a spectral delay deals with thousands of delay times and setting a reasonable feedback that works well with all of those delay times can be almost impossible. In this case it would be much more convenient to set the feedback in a way it takes the delay time into account. And that’s exactly what we are doing in our MagicDelay : when selecting the dynamic feedback mode, the feedback will be set in dB/s. So you are defining how fast your tail drops instead of a percentage to feed back into the input.

How would that look? Let’s examine the graph below:

Dynamic feedback, various delay times

Dynamic feedback, various delay times

(a) Input: To keep things visually simple we use a single sample impulse as input. Imagine a really short click sound.

(b) Input + Short Delay with 3 dB/s: This is the same delay time as (b) above, but this time we are using dynamic feedback.

(c) Input + Medium Delay with 3 dB/s: Same delay time as (c) above, but this time we are using dynamic feedback.

(d) Input + Long Delay with 3 dB/s: Same delay time as (d) above, but this time we are using dynamic feedback.

(e) Input + Very Long Delay with 3 dB/s: Same delay time as (e) above, but this time we are using dynamic feedback.

As you can see all the tail lengths match because we add the delay time to the equation and define the feedback in dB/s. This turns out to be really helpful when using spectral delays, and I would like to see this in conventional delays as well.

Conclusion

Adjusting the feedback in dB/s is a very useful technique, especially if you are dealing with thousands of modulated delays.

If you want to check for yourself just download and run the fully functional trial of the MagicDelay . It lets you switch between the dynamic and conventional feedback mode so you can easily compare those two. It is also mentioned in our Quickstart Guide - MagicDelay .