FM sound test

[chaosprint]

out: mix ~op1 ~op4 >> lpf 5000 1.0

~op1: sin ~op2 >> mul ~env_amp >> mul 0.3

~op2: sin ~op3 >> mul ##r()*1000# >> add ##r()*3000#

~op3: sin 2000 >> mul ##r()*1000# >> add ##r()*3000#

~op4: sin ~op5 >> mul ~env_amp >> mul 0.3

~op5: sin ~op6 >> mul ##r()*1000# >> add ##r()*3000#

~op6: sin 1000 >> mul ##r()*1000# >> add ##r()*3000#

~seq: speed 8.0 >> seq ~a

~a: choose 60 60 0

~env_amp: ~seq >> envperc 0.01 0.1

[chaosprint]

Everytime you update this you get some different timbre

[chaosprint]

out: mix ~op1 ~op4 >> lpf 3500 0.7

~op1: sin ~op2 >> mul ~env_amp >> mul 0.5

~op2: sin ~op3 >> mul 105.82071310647025 >> add 450.7365451914831

~op3: sin 1700 >> mul 350.3393009276775 >> add 800.3673627456516

~op4: sin ~op5 >> mul ~env_amp >> mul 0.2

~op5: sin ~op6 >> mul 7.904726478599743 >> add 550.5821910730873

~op6: sin 900 >> mul 90.51658867836811 >> add 170.61041392701827

~seq: speed 5.0 >> seq ~a

~a: choose 40 70 10

~env_amp: ~seq >> envperc 0.03 0.3

generated with chatgpt

[chaosprint]

out: mix ~op1 ~op2 ~op3 >> lpf 3500 0.7

~op1: sin 1500 >> mul ~env_amp >> mul 0.5

~op2: sin ~op4 >> mul ~env_amp >> mul 0.4

~op4: sin 800 >> mul 7.904726478599743 >> add 550.5821910730873

~op3: sin ~op5 >> mul ~env_amp >> mul 0.3

~op5: sin 700 >> mul 90.51658867836811 >> add 170.61041392701827

~seq: speed 16.0 >> seq ~a

~a: choose 40 60 10 20

~env_amp: ~seq >> envperc 0.01 0.08

[jarmitage]

Cool! Are you just doing few shot learning with GPT to achieve this?

[chaosprint]

just one shot gpt4, asking the model to find the pattern and create a new one.