Investigating Methodologies of Song Writing

horsetrainer

Below, I posted a link to a 'TED Talk" about "AIVA", an artificial Intelligence (computer) that writes music.

As I understand it. AIVA is a music writing program that analyzes the scores for hundreds of existing pieces of music, and looks for similar patterns among them. It is then able to design new music, by calculating new musical patterns based on rules of construction it "learns" from existing music.

I think this is interesting, and I think some of the AIVA music I've listened to does sounds good.

But I think there is more to producing spectacular new music than combining existing musical patterns. Although, I think it's obvious that throughout history, people preferred a style of popular music in any given era, and music within any given genre has similar methodologies of construction.

My own theory is the most successful music of any genre will be based around a simplistic melody of some kind, and other instrument parts only act to accentuate that melody in various ways.

This is the reason why I think most people initially "remember" music in terms of single notes of the predominate melody of the piece. Once it's remembered, then the mind can "hear" the entire piece playing in ones head. And if this theory is true, it might explain why some music is considered better than others. The music with the "catchy" melody is more likely to stick in the mind and be recalled.

But what then makes one melody catchy, and others not?

I personally think a catchy melody is one that elicits an emotion for the listener. It can have something to do with the quality of the vocals. I also believe the subconscious is very good at linking memories of familiar sounds, such as those found in nature, or those familiar sounds heard in day to day life.

Certain arrangements of musical pattern can have a quality that "reminds" the listener of those familiar "sounds from their world experience". The person may not be able to make a conscious connection, but the mind finds the music interesting because it is both new and familiar at the same time.

Perhaps familiarity is the basis by which this AI music generator is working? If it's deriving new music patterns from hundreds of pieces of "famous" music, perhaps the mind is bound to find familiarity within what such an AI creates?

sisterkate

@horsetrainer said:
My own theory is the most successful music of any genre will be based around a simplistic melody of some kind, and other instrument parts only act to accentuate that melody in various ways.

This is true in my experience my so called ‘successful’ songs were written when I only knew a few chords

But what then makes one melody catchy, and others not?

often the rhythm and the lyrics

Perhaps familiarity is the basis by which this AI music generator is working? If it's deriving new music patterns from hundreds of pieces of "famous" music

True for me I have learned lots of songs I liked before I wrote one of my own