Slashdot Mirror

Google is your friend.

I always wanted to have the ability to "hash" songs, and come up with an algorithm that would be robust enough to work across multiple codecs and encoding options, different (relative) normalizations, and maybe even be able to handle empty space at the beginning and/or end of the song.

It's been done. Here's a system where you can hum a tune and it tells you the song: http://www.musipedia.org/

Current systems are mostly based on pitch changes, so they aren't perfect (especially with the recycled slush turned out by low-grade high-visibility pop acts), and largely useless for rap, but they mostly work. There are numerous variations on the system, this is just one of the more significant ones that is publically availabel on the web.

I would think by making a hash based on values relative to sound signatures within the clip this might be possible, but I don't really know how this stuff works

What google is doing may or may not be related. They might instead be using a form of speech recognition technology, or a combination of both, or something else entirely.

... it's called Parsons Code (http://www.musipedia.org/pcnop.0.html). There is also a query by humming (http://www.musipedia.org/query_by_humming.0.html) .

L.

... it's called Parsons Code (http://www.musipedia.org/pcnop.0.html). There is also a query by humming (http://www.musipedia.org/query_by_humming.0.html) .

L.

Perhaps Google refused it because it has been done before.

5 minutes of research with their search engine might sometimes save a few weeks of development. If you are still keen to do your own engine (good on you!) then at least go the referenced site, where they explain that pitch is not necessary, only the sequence of whether the music goes up or down in frequency. You can whistle very imprecisely and it still works.

To find the pitch of what people are whistling, you can use the code from gstring, a simple guitar tuner, for example. The software is old, but that's because it has been working really well.

Good luck.

http://www.musipedia.org/whistle.0.html

Do I get a prize?

Here you go: http://www.musipedia.org/whistle.0.html

That already exists thanks to Alexa opening up their API: http://www.musipedia.org/

Musipedia operates alike, though it has a much larger musical repository (mostly classical or bizzare, the recent changes list shows a Super Mario and a Halo song). It's not perfect either, I tried the Canadian national anthem, and it ended up as 5th or so on the results.

People have been working on this problem for a long time. Check out papers published at ISMIR (International Symposium/Conference on Music Information Retrieval) for the academic work. The algorithms and technology are there, no question - the problem is the business model and the licensing. It seems simple, until you try to talk to record executives who don't want to license you access to their music catalog. It seems counterintuitive, since in theory it would make them more money, but they apparently don't see it that way.

Also, while it's technically feasible, it does get significantly more difficult computationally when you want to search an entire library of songs.

Here's an example of a free site that has the technology implemented, just without the database of popular songs (it works great for classical melodies).

Not trying to discourage you - the point being, either go into this because you think the tech is cool but don't expect to make money, or if your goal is to get rich, become a business major and be prepared to spend all of your time meeting with VC execs and recording industry leaders, rather than building cool tech.

Cubase gets me 80% of the way there quicker. But that last 20% is a nightmare (and needs redoing every time the music changes). Lilypond gets me 95% of the way there at reasonable speed; and the last 5% gets easier as you learn it.

What I have noticed is how much more natural its output is. Engraving isn't just a matter of putting the right notes in the right order on the right staves; there are an awful lot of subtleties in the layout and getting it to 'feel right' and be easily readable, making best use of the space. Most other packages (I've seen a lot from Cubase and Sibelius) are reasonably readable, but they feel robotic and unnatural. Lilypond isn't perfect, but it has far far better instincts about how to do layout.

And, like most programming languages, the more you use it, the easier it gets, and the more you can do with it!

It seems to be slowly supplanting ABC and other common formats as the format of choice for online music libraries. Musipedia uses it, for example.