Stephen Andersen, Aaron Schuman, Lee Ingram Jonathan Peard Evolution of Music Stephen Andersen, Aaron Schuman, Lee Ingram Jonathan Peard Everyone
Purpose of the Project Purpose This project will create songs using a genetic algorithm. It will generate new and unique songs every time it is run. Aaron
Uses of the Project Usages Production of novel musical compositions Use the critic algorithm for the analysis of music Inspiration for artists AaronAaron
Genetic Algorithm Overview Pick two random songs. Combine the songs in an operation called crossover into a new song. Small parts of the song will be randomly modified in another operation called mutation. Repeat. LeeLee
GA Step 1, Song Selection GA Step 1 1. Songs are randomly selected for propagation 2. The probability of a song being selected is based on its score
GA Step 2 GA Step 2, Crossover 1. A random crossover point is selected. 2. Elements from each song are added together based on the crossover point.
GA Step 3 GA Step 3, Mutation 1. Parameters relating to timing and frequency are randomly adjusted. 2. Whether a song is mutated or not is also randomly determined.
Critic Algorithm General Overview The critic looks at a song, evaluating it based on some hard coded musical principles. The algorithm looks at each sequence of notes in the song, and finds every pair of notes that play at the same time. If a rule is broken, demerits are given. The total count of demerits is returned to the composer. Jon
Musical Theory Musical Theory What makes a song good? Consonance Notes played together that are pleasant, are consonant. Tempo and Timing Notes sharing similar timing characteristics with their neighbours sound better. Phrasing Within an instrument’s track, one tonal scale is used and the two octave rule is obeyed. Resolution The beginning and ending of a track must have similar steps. Jonah Hex
How to Operate the User Interface Statistics are displayed and updated in real-time in the upper screen. Different statistics can be displayed with the tabs at the top of the screen. Real-time adjustments to the GA can be made on the center right. StephenStephen
Diversity Diversity The diversity of a song is a measurement of the difference between a single song and the average song. When active, a song’s chances of propagation are improved by its diversity. The use of diversity can be toggled with a button in the Diversity tab of the User Interface. (It is optional)
f(t) = A*sin(θat+sin(θbt)) Synthesizer Synthesizer The synthesizer is polyphonic, which means that there can be more than one instrument playing at a time. Instruments were created by tweaking a Frequency Modulator. f(t) = A*sin(θat+sin(θbt)) Stephen
Results Results The supervisor evaluates the provided music under a rigorous set of rules, and seems to favor classical music over more modern pieces. The composer creates songs with small sections that could be the basis for a piece of music. Both generated and translated songs have been played on the synthesizer successfully, with numerous instruments. A User Interface to display relevant and useful information, and adjust composition parameters, is operational. JonJon