0x0

So, my devember started with 4 days of delay because I was on vacation. Not a big deal.

On the first day I just opened Godot and started to poke around. Exploring the editor. A lot of strange things.

I began reading the Godot introduction guide and watched a video of introduction for Unity users. I sadly had a long and strong flu, so I lost a ton of days.

I aimed to implement a button that makes a text shake. This is what I've learned in the process.

A Scene is composed out of nodes

Node is the base class, everything else is deriving from that. It has nothing. Yes, no useless transform on empty nodes. Yay!

A node can have a single script attached to it.

To add functionalities to a node one has to create a new node class, extending from some other Node class. Usually this is not required though.

The normal thing is that you add nodes, maybe base ones, and add a script to them.

You can expose things in the inspector via the export keyword.

To get another node from the inspector you have to expose a NodePath and get the node like this: onready var my_node : MyClass = get_node( that_nodepath ) as MyClass. Node the onready keyword, that makes this execute after the nodepath and scene is ready.

You can save and reuse scripts, or you can leave them unnamed and internal to the scene. To name a class you have to use class_name MyClass. If you use class MyInternalClass it would be another class, interal to your file and not the class of the script. I've lost a good hour on this.

You can wire callbacks via the inspector, except they are called signals. You can use the "Node" tab, where you'll see all signals, divided by class hierarchy. If you double click one of them a chooser window will pop-up. If you have some signals connected on a node, a wifi-like icon will appear on the scene hierarchy next to it.

Ah, every node structure you can save as a resource, and it's still called a Scene. So a Scene of Godot is like a prefab in Unity. I don't think you can have variants though, but I'm not sure. You can have local modifications of a Scene assets inside another one. So maybe its feasible to get something similar working.

Here is the video of what I managed to get working today.

I’ve replaced the code that was generating the images you saw, with the functions I’ve extracted. Now I keep a float matrix and don’t work converting the initial spectrogram image. Having a single line on, it has no difference.

I’ve noticed that without the image normalization the recomputed spectrogram would look like the original, but with normalization it has black values on the line, so probably they are negative values, I’m not sure there can be negative values though.

The recomputed signal has frequency that seems double of the original from the inverse STFT.

TODOs:

In the inverse STFT I should use the window I used for the STFT.

The way I manage the time shift is wrong.

So, I hope you are not superstitious... :D

I’ve added multiplication by a Hamming window. It’s applied in the STFT from bottom left to bottom right.

I’ve made some progress on the STFT and iSTFT implementation. I had some sleepy-coding dumb bugs, even on the debug visualization. I still have to check if the STFT is correct. You can see the iSTFT of the top image below it, and it’s STFT at its right.

Thanks to Howard Antares, from the JUCE forum, I’ve found the terrible bug that was haunting me.

https://forum.juce.com/t/image-pixels-y-problem/43088/11?u=thegiallo

It was an uninitialized variable due to shadowing. I’m used to have all the warnings possible turned on, but in JUCE I didn’t think about setting them, so the compiler was silent and my mind continued to switch between the meanings of the `h` variable. It was both hue and height, at the same time, depending on the place. So strange.

So now I have this correct hue-saturation plot. Hurray!

I was curious about the pattern that was being created by the uninitialization, and in theory the variable would have the value it had the previous iteration, because of the stack. I tried to initialize it manually to that value and the pattern was still there! I’ve made a shadertoy so that the whole world can benefit from my sleepy coding beautiful discovery.

https://www.shadertoy.com/view/wsGfRy

I’ve now added the flags, but there it comes... no warning. I’ve used goldbolt.org to investigate the problem and discovered that `-flto` disables the warnings about uninitialized use. Also Fahien noticed that the variable pointed by the arrow should be the uninitialized variable and not the assigned one.

https://godbolt.org/z/qnd5xh

So... in these two days I’ve achieved... nothing. I did notice that the HSV image was wrong, no y changes. So I’ve investigated, and confirmed problems on y. I’ve tried many things with no result. As last resort I’ve asked on the JUCE forums. I’m really tempted to go back to my opengl code with hotreload and no portability and raw audio management. https://forum.juce.com/t/image-pixels-y-problem/43088

I’ve introduced modifying the image via the class BitmapData and checked that the result is the same as before.

I’ve been reading this tutorial and I’ve added an image with an ugly pixel by pixel rendering of a Hue Saturation gradient.

https://docs.juce.com/master/tutorial_simple_fft.html

The start is not certainly with a boost. I’m using a new library that even comes with a project management tool. I’ve downloaded the Juce build for linux and started reading the tutorials the website handily provides.

I encountered a problem with the build. I had to install libwebkit2gtk-4.0-dev because its absence was breaking the call to pkg-config and thus the build hadn’t all the system libs paths. Now it builds fine.

I normally use vim + YCM generating a compilation database via script, so right now I’m without YCM.

Being that my output was noisy and low in volume I tried to investigate the low-pass filter used in Pedal-Pi. It’s a Sallen & Key 3rd order. They link this pdf with all the calculations. After skimming through it, I googled for a calculator. I found this one http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm . I tried some values of the caps I have and finally found a combination with a cutoff at 20kHz~. I used 800Ohm and 10nF, with the couples of resistors and caps equal. To make the resistors I used a 470Ohm and a 330Ohm. After wiring the circuit up I only had 50Hz as output. It took me a lot to find out that I wired it exchanging the Vin+ with the Vin-. Probably it was oscillating, I don’t know.

Anyhow, this bug was lucky because I also found out the origin of the audio signal that could be heard with the program off. I got the polarity of the phone connector wrong. To me the best polarity was the one with the ground in tip and the channels back, in this way a mono and a stereo would have the ground in the same place and the positive polarity of the mono would be duplicated to the two stereo channels. Connecting a stereo jack to a mono output would work. I’m sure there’s a good reason why it is like it is, my knowledge is infinitesimal.

So, now my output is clearer and louder. I think I could solder the circuit on a prototype board. Oh, I also noticed that the coupling cap is shifting the output voltage, centering it near ground.

On the software side, I’ve implemented linear interpolation in the reproduction of the samples. Now it’s no more time-distorted.

Following the development during #devember 2019.

Reading the Pedal-Pi circuit I didn’t understand what C9 is doing. Thankfully with 20minutes of random googling I’ve found out what is it. It’s a coupling capacitor. It’s the exact opposite of the decoupling caps I’ve been using all over the circuit. Decoupling caps remove AC because they are in parallel with the circuit, coupling ones remove the DC because they are in series with the circuit. As simple as that. The oscilloscope shows that it cleans out some noise.

The green signal is the difference between the two, the yellow is after the cap and the red is before. The difference has the same frequency and inverted phase of the signal. I really don’t know why.

The output signal sounds very noisy. I noticed I can hear the input signal, very low, when the program isn’t running. Maybe some voltage is leaking? I hope the noise will reduce with a prototype board instead of a breadboard.