Merging direct signal with effected signal. Does a 1 or 2 sample delay make a difference?

Continuing the discussion from Ideas on the blend knob:

I’m taking this discussion to another thread since it’s off topic from the other conversation…

Assuming the internal sample rate is 48Khz, this means that one sample delay would put 24khz signal 180degrees out of phase. This is outside the audible spectrum but can have audible consequences to overtones. It’s like placing two mics a 1/4 inch difference in distance from the source. I know that when close micing a snare or guitar cabinet, this can cause audible issues.

Lets take this to two samples. Now it will be out of phase for 12khz. This is now in the audible spectrum.
It’s like placing two mics which are 1/2inch difference in distance from the source. Close micing a snare or guitar cabinet will definitely change the tone.

Is there a way I can experiment on the DUO?
Does a plugin turned off still delay the signal by one sample?
Or maybe use a gain plugin set to 0DB? Does the gain plugin delay the signal?

The MOD hard-bypass does not retain plugin latency. it just passes the signal through.

Since MOD 1.2 the mod supports so instead of hard-bypassing, the mod asks the plugin to bypass itself (click-free and retaining the latency if appropriate). This requires dedicated support by the plugin.

No. The majority of plugins on the MOD do not add any processing latency. Also most plugins that do add latency also modify the signal significantly that it likely won’t matter in a significant way.

Since there’s currently no audio delayline plugin available on the MOD, I don’t think so.

There is a discussion about this near the bottom of A stereo-width panner also sums signals and summing left + right of two spatially separated microphones can lead to similar effects (like the snare mic’ing you’ve mentioned).

Thanks for the education.

Interesting to learn how these things actually work…

Small correction there.
v1.2 does not support plugin-side bypass.
The new features for v1.2 have already been established and v1.2 will only receive bug fixes.

The support for plugin-side bypass (using LV2 enabled designation) is coming in v1.3.

Hi @Skydiver,
From the theoretical side, mixing the original signal with the same signal delayed of N samples is equivalent to applying a Finite Impulse Response filter to this signal for which the frequency response can be explicitly computed: it is equal to |H(f)|^2 = 2*(1 + cos(-2pifN)), where f is the reduced frequency (i.e. the frequency in Hertz divided by the sampling frequency).

  • For fs = 48kHz and N = 1, the system thus applies a 6dB gain @0Hz, a 3dB gain @12kHz and as you noticed the gain goes down to -infdB @24kHz (phase cancellation). So it does affect strongly high frequencies (which can be a limited problem for a dry electric guitar signal that doesn’t have much content above 10kHz, but can be clearly audible for signal with rich high frequency content).
  • For fs = 48kHz and N = 2, as you noticed the -infdB appears @12kHz. The gain is under 3dB for frequencies ranging from 6kHz to 18kHz which is 3dB under the maximum gain (that appear at 0Hz and 24kHz) and make a frequency hole that should be clearly audible (especially around 12kHz) for a very large class of musical signals.

I hope this help !

1 Like

Interesting… I’m not following the calculations but I think we both show that if there are two samples worth of delay there can be audible issues.

My thought is that there may not be much audible change but addressing these things can make the difference between a good piece of gear and an amazing piece of gear.

I’m not a programmer and can only give perspective from the physical world… but it’s something to consider. It sounds like, from what falkTX has mentioned, it will be addressed in an upcoming update. This is what makes this community great!

1 Like