I've been looking at http://www.youtube.com/watch?v=fCYpnVJsGuQ (the Black Mages version of Distant Worlds) lately, and been trying to transcribe it, for piano, synth, classical guitar, electric guitar, bass guitar, and percussion.

Problem is, there is a lot of polyphony, and I'm not too sure how to tackle it (especially listening to the piano harmony interspersed throughout the song). Is there an easier way to follow a harmony line than just "brute force?" Is there some program or something I can do to the mp3 file to make the harmony more evident?

Any perspective would be appreciated. Thanks!

Are you transcribing off of youtube or mp3? You'll generally need headphones for the job. Some headphones provide better instrument separation than others. You can EQ certain frequencies to bring out the harmonies more. Plus having ear training helps (recognizing dim, aug, min, maj, etc).

If there's a lot of poly (especially if different instruments are sharing notes, as often happens) you'll just have to do the best you can.

What might help is if you load the mp3 into Audacity (http://www.lmgtfy.com/?q=Audacity) and use 'Change Tempo' to slow it down (using 'Change Speed' will cause the pitch to shift at the same time).

You can also use a combination of high and low pass filters to isolate frequency bands, as brncao mentioned.

Ex. Low pass on 440 a few times, then follow it by a high pass of 220 (again, repeat a few times). That should get rid of a lot of stuff outside the 220-440 range. You can do it for any ranges as necessary. (440 Hz is the A above middle C, btw, while 220 is the A below. Just in cased you didn't happen to know.)

Another tactic is shifting the pitch. Maybe you want to raise it by an octave (12 halfsteps/semitones) to hear the bassline a bit more clearly.

Or lower it an octave to make out some really high notes, way up above the treble clef.

All of these modifiers are under the 'Effect' menu.

Audacity can help you greatly, even by using only a few of its features.


I kind of disagree with brncao about the usefulness of recognizing the chord quality. I think it's really being able to hear individual notes that is most important. Many of the chords won't conveniently be in root position, and they likely won't be in closed position either (with all the notes neatly adjacent to each other in a row), for that matter. Further, you will probably run into stuff like 6th chords, inverted 7ths & 9ths, etc. When it comes to that sort of thing, you can recognize majors and minors all you want, but it's likely you'll end up missing (leaving out) plenty of notes that way.

The bottom line is: Can you hear each note, and can you put it down on the exact spot that it should be on the staff for the particular instrument track you're on? (Not in the wrong octave, and not fitting neatly into some triad, but where it really should be.)

Another tactic is shifting the pitch. Maybe you want to raise it by an octave (12 halfsteps/semitones) to hear the bassline a bit more clearly.

Or lower it an octave to make out some really high notes, way up above the treble clef.
You should shift the pitch of the targeted instrument (especially the bass) to around 1khz-5khz (based from Fletcher-Munson curve or Equal-loudness contour). Remember, the more you shift the pitch (while preserving tempo) the more distorted it gets. At 12 semi-tones (octave) it's unbearable for me. However, If you want to raise the pitch without distortion, then the tempo must go up in proportion to pitch. I usually can catch stuff this way.


I kind of disagree with brncao about the usefulness of recognizing the chord quality. I think it's really being able to hear individual notes that is most important. Many of the chords won't conveniently be in root position, and they likely won't be in closed position either (with all the notes neatly adjacent to each other in a row), for that matter. Further, you will probably run into stuff like 6th chords, inverted 7ths & 9ths, etc. When it comes to that sort of thing, you can recognize majors and minors all you want, but it's likely you'll end up missing (leaving out) plenty of notes that way.

I disagree. Recognizing chord qualities is a shortcut, it narrows down the possibilities. Back in the old days when I started transcribing, I had no music theory, thus the dreadful of going through "trial and error". I don't want to say how many hours of my life I've wasted because of my lack of knowledge, but had I known a bit about chords, I would've gotten it done sooner. The song used only 3 chords: Am, E, and Dm. Knowing what chord I'm in is like a safety net, I can't go wrong if I use A,E, or C in an Am chord. Even knowing what key and mode it's in is also a safety net. It becomes a problem if you recognized a minor chord as a major chord or a minor scale as a major scale. Thus the safety net is broken. That's what I'm getting at.

That's as far as chord recognition goes. Everything else after that is what you've basically said. There are inversions, power chords (ei. A,E in an Am chord with no C), 4+ notes added to triads (is it F or F7?), which is very common in jazz, the spacing of notes, etc. This is where you have to get nit-picky with your ears to find, not recognize, whether the notes exist or not, or where they lie on the staff, which is what you're addressing.

Remember, the more you shift the pitch (while preserving tempo) the more distorted it gets.
You're right, I forgot to mention that I do also double-check with different variations on shifting, tempo, filters, etc. (For example, only 5 half-steps instead of 12) But I start with an octave since I've found it to be the easiest for me to work with in the bassline. And after it's all done, I always check against the original.


I disagree.
Well, i guess we'll have to agree to disagree.

Don't get me wrong. By no means am I advocating that people *not* advance their knowledge of chords and music theory, -however- I believe, as I said before, that picking out each individual pitch is the most important, *not* recognizing what the chord is. If the chord repeats later, it could very well have a variation in it. If you really too much on "oh, it sounds like the chord I heard earlier" you might miss the extra note or two added in (or subtracted) on the variation. That's my take on it anyway. Like I said,"agree to disagree".

I actually speed them up first. The songs themselves might not be slow, but they might be playing too slowly for me focus on the big picture. I usually notice a lot more (chord progressions, embellishments) than playing it slowly or regularly. After I've gotten all the information I can get from that, I listen to the songs at the regular speeds, and then I slow them down.

Yeah, everyone pretty much has their own way. It's good that we can all put out our suggestions though.

Hopefully, we can all learn new ways of approaching transcription.

Well, lots of input.

Ear training is not that helpful, especially for Uematsu's compositions; I can recognize the chords perfectly and everything, but there are just so many notes that fall outside the chord, or are in a strange rhythm.

I don't know much about tweaking EQ, but the problem is, piano harmony isn't much lower than the melody, and so tweaking the EQ shouldn't help that much, if there are other instruments playing. I think I'll try the frequency band isolation though (haven't use Audacity much); might help.

On another note, what about the technical specifications of headphones?

Well, lots of input.

Ear training is not that helpful, especially for Uematsu's compositions; I can recognize the chords perfectly and everything, but there are just so many notes that fall outside the chord, or are in a strange rhythm.



I disagree. Ear training are not just identifying chords. Transcribing is basically ear training. Also, if u can tell the chords and its different inversions, uematsu should be one of the easier ones. Remember that theres all sorts of chords and not just major minor triads but there are 7th, 9th, 11th, 13th, diminished, augmented, MjMn, half diminished...ect. In fact, I honestly believe that if you have a pair of good ears you would need very little understanding about music theory to transcribe any piece.

I disagree. Ear training are not just identifying chords. Transcribing is basically ear training. Also, if u can tell the chords and its different inversions, uematsu should be one of the easier ones. Remember that theres all sorts of chords and not just major minor triads but there are 7th, 9th, 11th, 13th, diminished, augmented, MjMn, half diminished...ect. In fact, I honestly believe that if you have a pair of good ears you would need very little understanding about music theory to transcribe any piece.
Agreed. Music existed long before standard music theory did. People still figured it out.