Tip:
Highlight text to annotate it
X
Welcome to the Evoltrio demo tutorial
Evoltrio is an application that generates music melodies with the use of evolutionary algorithms
This video will demonstrate the use and the various features of the program.
To generate a music melody, click the Evolve button
A window with the evolutionary process will appear and will disappear as soon the process is over.
The graph will show the improvement of the musical phrase for the ongoing iterations
Click the play button to listen the generated song
You can pause/resume or stop the playing melody
By clicking the Evolve button again the evolutionary process continues from where it stopped
From the drop-down you can evolve for a selected number of generations
The long horizontal lines mean that the evolutionary process converge to a local minimum
In other words after a certain number of generations the track can't improve any more
By listening to the result you can easily find out that the long-awaited track is not that great
You can restart the evolutionary process by clicking the reset button and then start the evolution all over again
It's important to restart the evolutionary process a few times to see the differences between them
The sound of the playback can vary depending on the operating system and the sound card
On my computer there is a sound glitch at the beginning of the playback almost every time :/
The Chromosome Chart tab shows how was assessed the best music chromosome from the eight fitness filters
Further information about the internals of the program will be available on the project homepage
The status bar shows the current iteration, the fitness assessment of the best chromosome
and a part of the best chromosome (in jfugue format)
Information tab shows the whole evolutionary process
On top of it there are the internal settings of the application (mostly for debugging use)
Just on top of the 1st iteration there are the enabled filters and the weights of them (more info on the site)
When the program converge the best chromosome remains the same,
as the process can't find a way to improve it further
A preset is a predefined set of settings that aims to imitate a musical style
When the user loads a preset, the evolutionary process is reset automatically,
Clicking Reset is not necessary at that point
The preset is displayed on the program window and currently is the smooth
The curious user can check the differences between the presets although it's not that important
The most important differences are those between the songs that program manages to generate
The Chromosome Chart shows that most probably the current song won't have notes out of scale
The Evolution Chart has noticeable differences between the different presets
Although there could be big differences in the Chromosome Chart from the different evolutionary runs
And that should have impact on the resulting music
Playback settings allow the user to change the speed and the instrument that plays the melody
The user can try 128 different instruments broken down into 16 categories
Some instrument sound much better than the others
e.g. the guitars sound very fake
Changing the Tempo can also differentiate the result quite a lot,
a very slow tempo can be boring with annoying long note holding
And a very fast tempo can "crash" the notes
Some instruments sound better with slow tempos and others sound better with fast
The user is encouraged to try all the available presets as the music assessment is subjective
Trying every preset for about 5 times is probably the most optimal,
except the fun_with_cowboys which is made for fun :)
Although to fine-tune a good result with a better instrument/tempo is more time demanding
Listening to several music songs can become quite annoying
but that's the trade-off when someone messes with experimental fields
which include a conjunction of art and science ...
The user can save the song in midi format
On of the most important aspect of the program is the rating submission
A real name is not necessary, on the contrary a username is suggested
The news of the project are going to be probably something like a monthly newsletter
The rating and the comments will help the assessment of the musical presets and
change some of those if most of the users dislike them
The rated songs will be posted alongside with the midi (available in mp3 format too)
Check the project homepage for further resources
You can browse the source code on the github repository
Happy music generation ...
and thank you for watching!