The Lounge is rated Safe For Work. If you're about to post something inappropriate for a shared office environment, then don't post it. No ads, no abuse, and no programming questions. Trolling, (political, climate, religious or whatever) will result in your account being removed.
You may link the two, but the world at large doesn't. E.g. Google "jackboots" and you will not find the word "Nazi" in the results page.
But whatever. Not sure there's any childishness involved, and even if it were intended as an insult, why not? Are we now in a world where we are not allowed to use insults? No-one on this site was being insulted, it seems to be just you who have taken offence for some reason, as a result of concocting some link between the word "jackboots" and "nazi", combined with a personal need to defend the EU.
(Oh, and yes I realise that Google is not a definitive source for everything, but it does tend to give some indication of common usage)
I may be totally wrong, but it seems that there should be a fairly good piece of software that can "listen" to a WAV or MP3 of a song, and transcribe (polyphonically) the various parts into staves of notes for that instrument.
Anyone know of such an application? Anyone aware of one being developed using machine learning?
You want software, that can hear a song, and transcribe each instrument, being able to recognise two different guitars, a guitar from a bass etc, in its right key, a key which is determined by the average of the notes played, and is not even always evident, into notes that depend on that key?
Easy in a three piece band perhaps, in an orchestra?
I think it's probably only working with videos so they can match the music to the motion of the instrument to some extent. There are also different tonal qualities, of course. But I wouldn't want to try differentiating between the first violin and the second.
And how do you determine the key? There are only 12 notes. Working out the key is very difficult.
Not really, as long as it's a "standard" key without accidentals. And, score notation-wise, there isn't difference between C Maj and A Min, etc.
We won't sit down.
We won't shut up.
We won't go quietly away.
I'm no musician but have a basic grasp. By your logic, it must therefore also be "impossible" for anyone to listen to a piece of music and transcribe it. I do know there are several tools (I've used some) that will take normal written music and re-write it in a different key. (The interface to do this was pretty complicated though, requiring the use of two keys pressed simultaneously; "shift" and either "up" or "down" as I recall) By implication, even if any automated tool gets the initial key "wrong" it would be a simple matter to adjust it into a more normally accepted key.
No, because (some) people (after many years in music) are very good at recognising scales, mood, and hence the key.
take normal written music and re-write it in a different key
This is simpler, much simpler. The notes and key are given. Trying to determine they key is difficult.
For example, the key of G is a happy key.
The key of E is moody and blue. A is simple, but fun.
Deep Purple used G a lot.
BLues uses E.
ACDC used A a lot.
How does a machine recognise mood? Emotion?
And then the issue of taking apart the sound and determining which instruments are playing which notes. Very hard to do for a human ear who knows music, and knows the make up of the band already.
For example it takes a good ear to recognise two guitars are playing a solo. It is often noticeable because the combination of notes is not physically possible to play. But you have to be a guitarist to recognise that.
A computer program, well, it can only do what it is told. It can not interpret for itself and decide.
Working out the key is a trivially simple task that the human can do, either as an input to the program or later when editing its output. In fact, since the computer already knows the *pitches* of every note, a good heuristic would be to go through all keys and see which produces the least amount of accidentals in the resulting score.
The really hard bit, as you pointed out first, is isolating the sounds of different instruments.
As you say humans can detect the key, if they are musicians, because of the mood, or feel of a song. How do you teach a computer to understand mood?
Indeed, this is difficult for computers, but it's easy for us and therefore not the problem we need them to solve -- accurate transcription (of heavily polyphonic music) is very hard for us and that's where we need help from the machine.
Last Visit: 20-Sep-20 14:19 Last Update: 20-Sep-20 14:19