How do I find out early whether I'm accidentally plagiarizing known songs?

Question

Similar to I accidentally composed a song that sounds like a famous song and Unintentional plagiarism while composing, now and then I accidentally write a song that turns out to already exist. For example, I wrote a piece for Big Band, and just before I finalized the arrangement I asked my girlfriend whether she wanted to listen to it, and she then remarked that the main groove was basically from Time of my Life (Dirty Dancing).

I'm not asking about what to do once I've found out. I'm already able to decide whether there is enough original material in the piece. My question is:

How can I find out whether my song, or important elements from it, already exists? Ideally in an efficient and reliable fashion.

I've tried all sorts of online music/melody recognition platforms. Half of them only work with a copy of the original song, and most "Hum to recognize" services seem to be unreliable. Furthermore, sometimes it's not the melody that I've taken, but a distinctive groove or instrumentation. I know of no services that claim to detect that. Finally, some melodies and songs I've accidentally plagiarized are actually unnamed pieces of film music or advertisement jingles, so they don't show up in some standard song databases.

Is there maybe a technique (analysing the song in some way to categorize it), some database that I can search semi-manually, a forum where I can ask other people (I imagine this to be a fun game potentially), or something like that? Or possibly important elements and aspects (chord progressions, melodies, rhythms, ...) that I can extract and search for these in isolation?

My question is not about the legal aspects of plagiarism. It's rather that I've found that I'm much more enthusiastic about those pieces of mine that are more original, because I feel prouder with them and see myself more in them. I can define my personal style with them. Finding out that my piece already largely exists usually takes a lot of motivation away from working on it, so I'd rather like to find out early.

Answer 1

Unfortunately, I think the magic solution you're hoping for doesn't exist. There are various places that musical works are "compiled," like hymnals with hymns listed by their tune, but for the most part, there's no "database of chord changes, grooves, or melodies," just because indexing it would be so hard. Probably the most efficient way to do what you're hoping for is simply crowdsourcing and asking folks "what does this remind you of?" You could try to rely on people you know have a broad knowledge, especially in the targeted genre.

One important question is: So what? I mean, as long as it's not true (legal) infringement. Obviously there's also a gray area where you might not get sued, but others might look down on you as having "ripped off" another work in a broad sense. But beyond that zone is a very broad area where most music is legitimately influenced by other music. Music that doesn't sound like anything else is pretty rare, and of very niche appeal (Frank Zappa, anyone?). Genres—the fact that they exist at all—are all about commonalities. Most funk songs share similar "grooves," as do most reggae songs and for that matter Viennese waltzes and baroque minuets. Maybe there's nothing wrong with stumbling onto a "sound-alike" and then altering it enough to be safe.

Answer 2

Let’s first talk about plagiarizing. Two pieces sounding similar is not plagiarizing. Plagiarizing is (intentionally) using work of a different person and passing it off as your own. You writing something that by chance sounds like something else is not plagiarizing (and all of it is original material, it just coincides with a different piece).

And it is bound to happen, and has been happening all the time. Especially with music of very low complexity such as popular music and using a common musical language there really are limited amounts of possibilities, and thus it is quite hard to come up with something totally dissimilar to anything that has been done before.

Think of it like this: If music were books, then popular music would be a single sentence of limited length. And assuming common Languages there are only so many such sentences that actually make sense.

What can you do to avoid this? Write more complex music, for example. Go into the direction of complex jazz or contemporary classical or something. Or change the language. Use uncommon musical paradigms.

Both are most likely things you do not want to do.

Now to answer your actual question: No, I do not think there are (openly accessible) databases for this kind of thing, at least not to the extent you’d need.

And yes, of course you can break down similarities to theoretical things, such as:

• Chord progressions
• Rhythmic patterns
• Structure
• Structure of melody (mons/pons/fons and stuff)
• Key notes of melody
• ...

Answer 3

Have you tried uploading your song to YouTube to see if you get a copyright strike? I don't have direct experience of this, but I hear from others that the algorithm of YouTube seems to be quite sensitive and will sometimes/often detect copyrighted work even when the uploaded version is played in a different arrangement on different instruments to the original copyrighted version. Perhaps it's worth a try anyway. Let us know if you discover anything interesting.