Log In  


Cart #charlie4-3 | 2019-04-24 | Code ▽ | Embed ▽ | License: CC4-BY-NC-SA
8

Charlie is back from the research lab! Now he can lock the brain into keys and scales. He will still have to teach his neural network (music brain) how to make musical sounds.

When he starts the machine for the first time, it will make a random sound or maybe nothing at all. To hear the sound of this brain again he can press the green PLAY button in the upper left. To reward the machine for playing something sort of musical, he can press the red LEARN button. This will modify the brain and a new sound will play, usually similar to the first but not always. If the new sound is still "musical," he can press LEARN again to get more creative. To save this brain to the cartdata, he can press the gray SAVE button. To randomly reset the brain's concept of music, he can press the blue RESET button.

The current sequence is visualized at the bottom of the screen. If Charlie approaches it, controls will slide up to change the behavior of the current sequence.

Charlie can control the sequence's speed and length by changing the HEART (low for fast) and HOURGLASS values. The two hourglass values represent the beginning and end of the loop, which is highlighted in red on the sequence.

Charlie has researched the twelve tones and developed a set of keys and scales that can be selected by changing the MUSIC NOTE value. Changing this value will not have an immediate effect, rather, all notes produced afterwards will be in the selected key.

The SPIRAL value is a chaos control that affects the randomness of the brain's impulses on the next play (no immediate effect).

Charlie can add and remove sequences with the +/- control in the bottom left. More than 4 creates 'custom instruments' which get used by the current sequence.

If the brain is not making musical sounds, he can modify any node directly by moving on top of it, pushing the value up or down with the buttons. If he presses both buttons on a node, it will change randomly. The weights of the nodes and edges are represented by color patterns. The first column of nodes are input nodes associated with instruments, pitches, envelopes, and volumes respectively. Modify them and press play to get a new sound from the machine using the current brain, without changing any connections. The middle columns are learning nodes. Modify them and press learn for chaotic, unpredictable behavior. The right column are output nodes mirroring the input nodes. Modify these and press learn to train the brain to produce those outputs when given the current inputs. (left column)

The brain is answering a simple question, "what note should come next?" The input (left) being the 'current' note and the output (right) being the next note. When Charlie presses play, the brain answers that question for what is currently on the left, producing an output. It takes that output, and puts it back in as input, and repeats to generate a sequence of notes. The sequence of notes that it has generated are visualized at the bottom of the screen and highlighted as they are played. If Charlie moves on top of it he can access a control to add and remove sequences.

Good luck have fun!

8


Charlie's been tweaking his machine. Now you can control sequence speed, start and end and immediately affect the sound playing. In the pause menu there is FOCUS and MEDITATE to hide the background or hide the UI. Also you can enable mouse controls (kinda buggy).


bumping this thread after a long hiatus!! please help me test the new version


It's uh... interesting, for sure. I just don't know what I'm looking at or doing xD

I mean it's a neural network that generates music, I get that. But the 'user interface' is more artsy than functional :)


Thanks for trying it out, @Xii ! I'm sorry it's so confusing, designing an interface for this is a challenge, keeping it fun while also exposing the internal mechanisms.

I should mention there's more info on my itch.io page https://notb.itch.io/Charlie

How I Play: Randomly mash Reset/Learn till I get something that's kind of interesting (lots of different colors in the sequence at the bottom, rather than all same).

I'll then try tweaking the chaos at the bottom right, and doing Learn some more until I'm satisfied.

Then I start actually building music by picking tempo and loop lengths and adding more sequence tracks.

If you're happy with your brain, I recommend adding more tracks up to the maximum, to fill in the "custom instruments" with your own music. You can then remove the extra tracks and the custom instruments will remain. (This is an obscure feature that I prefilled with some defaults)

I generally don't mess with the brain cells directly, although it does have interesting effects, starting out this way is not as effective. The network is more a visualizer of its thoughts, so you can tell if the brain is thinking something or nothing.


wow!!!! impressive artwork!!



[Please log in to post a comment]