adding comments to filterpansingle

[jclaveau]

Hi @daschuer ,

Thanks a lot for your work, I read it carefully despite my little c++ skills. I don't really know if I can call it a review considering how much I still have to learn but there is at least one thing I can share : My fresh look as a beginner. So I added as many comments I needed to understand what you did.

Could you tell me if there is any misunderstanding in my descriptions before I merge all what you've done?

More, thanks a lot for this example of a ramping applied sample by sample, it would be really more flexible than buffer than buffer. Don't you think I should use this pattern to wrap the calculations made in the autopan effect (https://github.com/demos/mixxx/blob/pan_effect_stereo/src/effects/native/autopaneffect.cpp#L164-L188) in a separated class (could be useful in a true pan knob/preference)?

There is one question remaining for me, it concerns the naming :
"Enginefilterpansingle" doesn't talk about delay even if it's mainly what it does, wouldn't "delaypan" be clearer? More, if i move the linked piece of code into another class shouldn't I call it enginefilterpansingle?

Sorry for the time I needed to response and thanks a lot for this "lesson" :)

[daschuer]

The spreading into classes is Ok for me now. If one likes to reuse the code, it is a good time to split the code up than as needed. I have also thought about adding the gain value to the Enginefilterpansingle ...

Enginefilterpansingle Yes, the name is stupid and Enginefilterpan is now unused.
Feel free to pick a better name.

[jclaveau]

• Notes addressed (timeBetweenFullSamplesLeft is really self explaining 👍 )
• Concerning the naming of the class, i prefer to wait having a real knowledge of the idiom before doing this kind of things.
• Would you have something to advise me to learn differences between ramping, different possible linear interpolations or the threshold I added on my gs.frac value?

Thanks in advance and for the fast reply.

[daschuer]

Ramping: Inside Mixxx we use this value to go from one state to an other. For example if you turn the Gain knob, Mixxx samples the knob position lets say 0.50 0.55 0.60. If we apply these abrupt gain change to a audio signal, it adds hear-able square wave noise to the signal. To avoid this, we change the gain for one buffer lengths form 0.5 to 0.55 and after sampling 0.60 from 0.55 to 0.60.
This still applies noise to the signal, but this time triangular noise, which is much harder to hear.
The ideal solution would use a form of a han window or similar, but that is hard to calculate.

In this case, we just try to guess the signal voltage between two samples buy using a straight curve between them.

In the old solution we have used ramping, because we have faded out the samples with the old delay, and faded in the samples with the new delay.

or the threshold I added on my gs.frac value

What do you mean here?

[daschuer]

Thank you for your improvements!