Hacker News Reader

That would only work for songs that always stay in key, many don't. Normalizing to C is a better approach.

The numbers in Roman numeral analysis are relative to the current key of the song. It works fine for songs that modulate.

Assuming they remain diatonic, much music doesn't.

Why do you need that assumption? I think Roman numerals allow for non-diatonic chords like a i diminished.

Roman numerals still presume you're playing on a harmonized scale in a key; not all music fits that as many rock musicians don't know what a key is and pick chords randomly by sound, not by knowing anything about music theory.

The parent to this comment said

> It works fine for songs that modulate.

By definition, modulation is the act of not remaining diatonic. Modulation = changing keys. Diatonic = staying within key.

So, no, we don't need to assume anything about diatonicity in order to successfully do harmonic analysis using Roman numerals.

I always thought of diatonic as playing within a harmonized major/minor scale, even if you change keys. What I'm talking about is music that doesn't obey the rules at all, doesn't have a key, doesn't change keys, it's just random chords thrown together because the artist thought they sounded good. Changing to another key is still playing in a key. Many rock musicians wouldn't know a key if it bit them in the ass.

Reply on news.ycombinator.com