hidden markov model
A probabilistic state machine populated with note values.
For each note in each sequence build a hash-table of the counts next notes found
var data = {
"C60": {
"D62": 700,
"E64": 300
}
}
Then normalize each element in the hash-table to get the probabilities.
var data = {
"C60": {
"D62": 0.7,
"E64": 0.3
}
}
So if you get the note C60
there is a 70 percent chance of going to D62
.