h a l f b a k e r yLeft for Bread
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
NOT software that recognizes tunes.
Software that creates sheet music for any music it hears, from basic melodies to orchestral performances
Early versions
http://www.digital-...igital-ear/info.htm
[theircompetitor, Oct 04 2004, last modified Oct 05 2004]
Cognitone
http://www.cogniton...nav/intro/page.stml
[pashute, Aug 16 2011]
[link]
|
| |
Would work with one channel only, once music is mixed, it's like trying to seperate individual colours from mixed paint. But nice idea though... |
|
| |
There is software available that writes down anything you play on a keyboard. This sounds v.useful except unless you play metronome regular it writes hemedemisemiquavers all over the place. Neither can it cope with even a bit of syncopation. Tell me how your version doesn't do this and I might remove my fishbone. |
|
| |
what [simonj] says. to recognise an individual note when only one note is playing would be relatively simple - it would be a case of of interpretting the note frequency into a note on the stave *. When you mix notes, things would get much more complicated |
|
| |
* you'd also have to have a function to define a note before hand. For example, UK orchestras tune to Middle A 440Hz, whereas I believe European orchestras tune to A 444Hz. |
|
| |
hmm, I'm sure it's complicated, but we're able to actually process speech, which is much more complicated -- you're talking about frequency detections here, after all. I don't see why it would be impossible. |
|
| |
Half Baked - someone did this before I thought (I can't find it in a quick search, though). If not, it does seem like a good idea. |
|
| |
I think there is MIDI software that can recognise the basic timing and "correct" your slight timing errors. Of course, if you are too far out, then it will correct in the wrong direction.
For recognising notes from an orchestra: I think the problem might be in the Fast Fourier Transform of the sound. Most instruments have a big series of even and odd harmonics and these will superimpose onto another instrument. Taking the fundamental frequency of one instrument is relatively easy, but if it has an adjacent harmonic from another instrument, then things get difficult.
It's a nice idea, but until you explain how it can be done - fish, sorry. |
|
| |
In MIDI sequencers the function that corrects slight timing errors is usually called "quantization." In sonic software there are autotuners for pitch correction and, also, there are systems to correct the tempo during recording ([waugsqueke] told me about the latter but I don't know what they're called). |
|
| |
sounds too much like magic. |
|
| |
Why would this be magic? A human can do it, and a computer can already do a simple melody. |
|
| |
Recognizing different instruments by tembre and pitch or even by sampling is certainly possible. Recognizing pitch and volume is also possible. |
|
| |
A non-trivial computing problem, but not magic. |
|
| |
I have a friend who got his Ph.D by trying to bake this. |
|
| |
Yes, it involves lots of FFTs, but that's just the start of it. |
|
| |
To summarise: given a piece of music, you can tweak the algorithm to make it work well. The hard bit is making it adjust itself so that it works well with arbitrary pieces of music. |
|
| |