I'd go for hands on combined with a good text book.
Pick a device you are interested in, then purchase a kit to build it from. I suggest a kit rather than using a schematic from the internet because a kit will have (most likely) been reviewed and corrected. A schematic from the internet may have been designed and built once - there may be errors or weak spots in it that weren't discovered. I say that out of experience. I post most of the things I build on the internet, and I know that I make mistakes.
Purchase a good soldering iron and simple tools. It doesn't take a lot. A pair of needle nose pliers, a pair of tweezers, a desoldering bulb or pump. Make sure to get a soldering iron with a temperature regulator. Don't start with an unregulated iron or one that regulates the power. You want one that controls the temperature. (I started out as a kid with an unregulated iron. You can learn to use an unregulated iron well, but you will damage things until you get the hang of it.)
Assemble the kit following the directions it comes with.
A good kit will include a schematic diagram as well as the assembly instructions. It should also include a description of how the device functions.
Assembling the kit will give you a view of what it takes to make a workable device - how things have to be arranged to make a practical, usable device. Where controls go, where inputs and outputs go, how to connect power, etc. Pay attention to where things are on the PCB in relation to where they are on the device housing.
Go through the schematic and the functional description from the kit and identify the functional blocks of the circuit - things like amplifiers, buffers, filters, clippers (if you are building a fuzz box,) etc. Look them up in a good text book (The Art of Electronics is a common recommendation) and read up on how the circuit blocks function and the theory behind them.
Learn by observation how the mechanical parts work. Learn by comparison with a good text how the circuit works and how it was designed.
Once you've got a little experience, pick out a device you'd like to have. Break its function down into individual blocks and look up how they work. Design each block and join them together to make the device. Build it, try it, find errors, correct the design, make it work.
Signal processing is more a question of programming than hardware. There are many ways to get into it.
A simple way to start is to get a copy of GNU Radio. Despite the name, GNU Radio is not specific to radio signals. It is perfectly happy to work with audio signals - it can use the soundcard in a PC or a Raspberry Pi for signal input and output.
You can design your signal processing chain using the graphical GNU Radio Companion or you can write Python code to chain together the various GNU Radio functions. You can also combine the two approaches - the "humdrum" setup of input and output in GRC together with a module in Python to carry out whatever complicated processing steps you'd like.
Develop the functions you'd like to carry out using GNU Radio. Once you know what needs to be done, you can translate them to another language for standalone use or more performance. You may also consider porting it to a microcontroller and optimizing it to run within that more restricted environment.
Whatever approach you choose, keep in mind that it will be a long process. There is a lot to learn, and a lot of interesting things to build and do. It can be very rewarding - and very frustrating.
When you try something new, give it your best shot using what you already know and what information you can find on your own. When you get stuck, ask for help. Nobody can do or know everything alone - there's simply too much out there. Ask for help. It is almost certain that someone has been in a similar spot and can give you suggestions and assistance.
Many people will suggest using simulators to learn about electronic circuits. Pretty much everyone uses them at some point.
Keep in mind that simulators aren't perfect and don't always reflect reality.
You need to have some knowledge of what you expect the circuit to do so that you can judge whether or not the simulator is telling you the truth.
It often helps to have built a real (though simplified) version of a circuit before you try to simulate a more complicated version.
One of the most useful tools for electronics is an oscilloscope.
As a kid, I owned a functional oscilloscope before I owned a good multimeter. I bought the 'scope at a garage sale for just a few dollars. It wasn't accurate, but it let me see what was really going on in the circuits I built.
I find that it is often really useful to trace a signal through a circuit to see what is going on. Linear regulator putting out the wrong voltage? Oh, the output is oscillating. You can't (directly) find that with a multimeter.
It doesn't have to be a good scope, or a new scope. It just needs to be functional. Mostly you don't need a precise measurement when trouble shooting. If your switching regulator is running at 101KHz instead of the intended 100KHz, well, so what. It's more important to know that it is switching at all and that it isn't switching at a completely wrong frequency.
A used scope that you can afford now is better than a brand new one that you have to save up for a year or two to buy. It won't have all the snazzy functions, but it will give you insight now.
The inexpensive little scopes that you can buy on Amazon and other places generally aren't worth it. The input range (voltage) is too low and the bandwidth (frequency range) is too low to really be useful.
A used professional scope that is twenty or thirty years old shouldn't cost much more than the new cheap junk but it will serve you better.
