Is there any tool anywhere helping to create a score from an mp3 song?

• Aug 26, 2019 - 22:51

I don't mean in MuseScore, I mean anywhere. I found mp3 to midi converters (like and it works, but the result (attached) is, of course, a mess in look and sound, while the main theme still can be heard. Is there any way to streamline it?

Attachment Size
arlekindraft.midi 37.91 KB


In reply to by etsenberg

There have been discussions on this forum before on this subject, and the short answer to your question is apparently “no”. There have been attempts but the technology required is still in its infancy and doesn’t provide useful results. Although the simpler the music the more likely something usable will result. AI is not yet smart enough to distinguish a french horn from a trombone from a bass guitar from a tuba from a ‘cello, or if a given tone is in the melody, the harmony, or is a harmonic of a fundamental frequency.

In reply to by marty strasinger

My task is the most complex, it's creating notes from human voice. At least creating the proper rhythm, e.g. a note (even unpitched percussion) for every major peak in the mp3, would be a great help... I tried to do it manually by measuring the times between peaks, but they don't match the standard notes length, and when I adjusted them, roundoff errors lead to a way different tune. If done automatically, tempo change could be used for every such note,

In reply to by marty strasinger

For this monophonic melody AnthemScore should easily make a good job.
I only tried the demo with good results (classical guitar or Singing or small ensemble) but the 30 days are over and so I can't post a result from your singing.
Often people also suggest Transcribe!
here's a discussion

As it is not too often I am transcribing pieces (if I had to do it odten I would buy AnthemScore) Mostly I use Reaper to slow down a recording and do it all by ear.

Regardless what Tool you would use you would have to have a musical foundation for creating a score out of it.
At first: I really like your created youtube-Version. It almost has nothing to do with the original but sound interesting in its own way :-)
If you would eg. use AnthemScore it would be better to sing the melody without words because that just makes things more difficult for any software and no software is able to also recognize the words anyway.
But despite of any tool, you clearly have an advantage when you already get aware of the Key Signature and Time Signature of your piece to transcribe. So in your example it sounds like B-minor in 3/4 with an upbeat of 1/4.
Here is an MS-Score of my draft of the melody of your Singing transcribed by my Ear:

Attachment Size
Arlekin.mscz 6.48 KB

In reply to by musikai

AnthemScore is not XP compatible :(
OK, back to manual conversion. I'm impressed how fast and easy you did it with my first stanza, I made only 2 adjustments. What surprised me most is that you managed to do it with 1/4 and 1/2's and to fit simple 3/4 measure while I, after analyzing length of sounds in my mp3, used all kinds of notes from 1/8 to whole, with augmentations, and came to exotic 21/4 measures to avoid splitting a note between measures - and still resulted far from the original rhythmic pattern. Can you give me any advice for finding the correct time signature? Something more mathematical than just "you should feel it" ;)

In reply to by etsenberg

Thank you , Jojo-Schmitz , for converting the file. This is the first time I used MS ver3 as I was so happy it now also exists in 32bit. Didn't know about the non-backwards-compatibility but do now!

thank you for telling your experiences with my transcription!! It is a nice melody. Please tell me is this a traditional song or have you composed it? Does there also exist an english or even german version of the Title and lyrics? Or can you post the original lyrics so I could let it translate it?
I'm asking because I'm a bit inspired to create an easy Classical guitar piece out of it.

Yes, AnthemScore is 64bit only, this is another reason why I didn't aleady buy it, as I'm mainly still working on an old weak 32bit notebook with Win7. But if you have 64bit XP you still could try if AnthemScore would work on it.
Now to your questions:
To be able to create a Score of a Melody you must be able to read notation and sing or play from reading the notes. Only then you can go the other direction.
When I transcribed your piece finding the rhythm was not hard because I often played similar pieces and so it was easy to recognize. But this doesn't help you much.
Essentially the first thing is to find the pulse. Listening to any piece you try to tap along with it with the fingertips or hand or clapping hands until you have a steady beat: t - t - t - t - t - t - t - t...
Then you try to find the heavy and light beats. Find the "1" to find out if it is eg. a 3/4 or 4/4 Time signature:
3/4: T - t - t - T - t - t - T - t...
4/4: T - t - t - t - T - t - t - t...
So you might find out that your piece doesn't start on a heavy but on a light beat->It's an upbeat.
Additionally to that the intervall from your 1st note to the second note is a 4th up - like many many pieces start with an upbeat and the melody goes up a 4th. The upper note is the tonic of your piece and the lower note is the 5th note of the scale (but 1 octave lower). So your piece starts with this typical sequence: The 5th note of the scale in the upbeat-note, goes up a 4th to the tonic (1st note of a scale) in the first heavy beat.

So this is part of finding the Key Signature of your piece. First you try to find the tonic note. Mostly it is the ending note of a piece or the first note (after the upbeat). Then you try to find out if it is a major or minor scale etc.

If you don't already play an instrument I recommend learning to sing or play from notes. This is the only way to be able to create a score. When you learn how to sing or play major scales, minor scales and know how they are built, when you learn how sing intervalls and find the correct note then you have much more skills to do the same in the other direction.

In reply to by musikai

Thank you for your detailed response! I am very pleased that you liked my song because I wanted to ask you to transcribe the rest 4 stanza (the 6th is the same as the 1st), and as you want to play it yourself, there will be no problem ;) This song is completely mine in terms of music and 1/3 mine in terms of lyrics (based on the poem by Joseph Brodsky - I hope you heard about him, he was a refugee from Russia to the USA (like me) and Nober prize winner (unlike me yet ;)), but I rewrote the original rather significantly. making it even more dismal). I've translated my version of lyrics for you (and everyone else):

Arlequin's song
(I know they spell "Harlequin" in English, but the original Italian name has no "h", and I hate distorting names in translation)

by Joseph Brodsky and George Right

Through all kind of land I am drawing my show,
And what did I see, living long enough?
A piece of flash is plodding below,
A piece of metal is flying above.

From age to age we are same way bound:
All living persons will lie underground,
Unhappy and happy,
In love and in hate -
Yet countless millions have met their fate.

Some men say - try
To comfort yourself,
And others cry -
For them you should die.
For some - due to age,
For some - due to rage,
For some - due to love, or books on the shelf.

But will you be good or will you be not,
No one is delivered from everyone's lot,
Whatever you wish, whomever you trust,
But everything living will vanish at last.

Some men say - you go to paradise,
And others say - your mark stays when you're gone,
But all this are only consoling lies!
And what's the way out? And there is just none.

Through shadowy land I am drawing my show,
And all what I saw, living long enough -
A piece of flash is plodding below,
A piece of metal is flying above...

English is not my first language, so in case of any weird phrasing don't hesitate to tell me. BTW, is your first language English or German? I'm asking because I need someone to sing meine deutsche Lieder ;) I'll never dare to do it myself due to my terrible accent. I'm not a good singer even in my native language anyway. While I made up some songs since my childhood, it's hard to imagine a person less similar to a composer than me ;) I never played any instrument (tried a guitar as a teen but didn't get enough patience to learn), I cannot recognize any note neither by ear (what note it is) nor by eye (how it should sound), and when I started using Musescore several months ago, my knowledge in music theory was absolute zero (it is a bit better now. but not much). However, I wrote 13 pieces since then (or 14 if we consider vocal and instrumental Arlequin song as 2 different ones ;)), and I like how they sound. Your can listen to them (and see the scores) here: They all are songs, most on my lyrics, some on others.

So, I'll be glad if you complete the transcription of vocal Arlequin ;) As I've said, I made just minor changes in your version of the first stanza (attached). Arlekin_dr.pdf

In reply to by etsenberg

Thank you for the lyrics! Interesting poem!
Oh, no, you don't want to hear me sing your german songs :-)

Because you asked so nicely here is the transcription of the whole piece as I hear it and hope it fits mostly to the russian lyrics.
I watched your songs and the notation. It's quite a wonder that you were able at all to create a notation as you don't have a musical education. But I recommend to look for some musicians around you to help you with the transcriptions to get them musically correct.

Attachment Size
Arlekin_Verses_MS232.mscz 24.71 KB

In reply to by musikai

Thank you very much! Sounds as it should, I am astonished that you got it right from my imperfect singing ( only in some places 1/8+1/8 should be turned into 1/4 or vice versa). And what's your name (or nickname if you prefer it) to attribute you properly when I put it online?
If you record your version with English lyrics, please send me a link/mp3.
Of course I am able to write notations because MuseScore helps me. After I find the proper rhythm, I just move the notes up and down and try how they sound until I'm satisfied. But all these A, Bm, F#(m) are still kind of Greek to me. I've read (just days ago) what major and minor scales are, but didn't get a good understanding of it yet. The same note can belong to many major and minor scales simultaneously, doesn't it? So what's the sense in these symbols? All notes within a single measure should belong to a single scale? But I can change the borders of the measure at any time, like split 4/4 into 2/4 or vice versa, and the melody will be the same - only the notation will look different. And if there is no fixed borders for keeping the same scale, again, what's the sense in it?
Unfortunately, I know no musicians, so internet is the only source of info for me.

In reply to by etsenberg

Hi etsenberg,
glad you like it. No need to mention me but I do appreciate it of course. In the attached score you can find my name.
Ah, the Text above are chord symbols for accompaniment. Don't let you confuse about that.
Your whole piece is in the Key Signature B-minor. So you sing a minor scale based on the fundamental note B.
This is why there are 2# at the start of the piece as the notes of the natural B-minor scale are:
b c# d e f# g a
If you would sing the whole piece one note lower then the Key Signature would be A-minor. The Key Signature A-Minor wouldn't need any accidentals as the notes of the natural A-minor scale are:
a b c d e f g
So besides finding the correct time signature it is also important to find the correct Key Signature for correct notation.
But there is so much to explain that it is hard to do all here. Perhaps a good idea for a singer would be to have a look at solfeggio or the similar approach of Kodaly.
If I find a good website for music self education I will post it here. (or anybody here)

So the chord symbols above are a suggestion for musicians for playing an accompaniment. In the attached score you can find some very very simple examples how a guitar player could play the chord symbols. These are just some standard picking patterns but at least you can now hear the chords. For the fun of it a little intro came to my mind (8 bars at the start).

Attachment Size
Arlekin_Verses_MS232_guitarchords.mscz 28.89 KB

In reply to by musikai

Thank you very much! And this is the final piano version (I split or joined some notes where appropriate for Russian lyrics, and increased tempo).
I have 2 questions. The small is, why did you turn the 1st measure into 1/4 and what is the mark (like bold 1) over it?
The general (unrelated to a certain piece) is, why do we have 7 notes with 2 phase shifts (half-tones between E and F, B and C) instead of having much more logical 6 notes with a full tone between each 2 sequential ones? We would have all the same sounds, just F would become E#, F# would be F and so on up to A - A# - C (no B), and it would be more logical and simpler, especially for transposing and chords. I've read articles at different musical websites and asked people in internet, but couldn't get the answer. Everywhere it is given as just a fact w/o explanation of the reason.

Attachment Size
Arlekin_piano.mscz 24.18 KB

In reply to by etsenberg

oh I like that very much! Your voice (and accent) play good together with the poem. Little question: Is "Flash" really "Flash", or isn't it "Flesh"?
You asked some questions.
The bold 1 (and 2,3,4,5,6) were only used as marker for the different stanzas for better orientation.
If a piece would start with rests (like your piece would start with 2 1/4 rests) then one can just omit them. So the first bar, which is not fully filled is called a "pickup bar".

The other question of you that sounds like an easy question at first sight is not so easy to answer. The way you sing without having an academic musical education is nontheless based on the tradition of our culture which uses heptatonic (7-notes) scales. These scales use 7 different notes with specific frequenzies or relative ratios. The major scale and minor scales of the western music are only 2 specific scales out of a huge number of other possibilities. I also was surprised what is out there in the world when I read this article:
As you found out the major scale uses half and whole steps. This is called a diatonic scale:
Starting from that you can perhaps dive more into the topic.

Your idea to just use whole steps is a nice idea. It is very logic, but you will see that it will sound very strange.
It is a 6-tone Scale and really exists as "Whole Tone Scale". Composers made use of that in their Compositions.
Please have a look at the attached Score, where you can listen to a comparison between a Major Scale and a Whole-Tone-Scale and try to sing along with it :-)

Attachment Size
Major Scale vs Whole Tone Scale.mscz 13.96 KB

In reply to by musikai

Flesh, of course. That's a professional deformation of a computer user ;) But the difference is not audible, so I corrected only the text but not the sound file. Here is the complete video:

Yes, there are many different scales, not all of them diatonic, but most of them still use 7 notes, and I cannot understand where this number came from. The only point I found is that peoples with tonic languages tend to use pentatonic while non-tonic-speakers use heptatonic. But that's not the answer, why. I understand that using whole tone scale ONLY would limit the composers capabilities, but. as I've said, with alterations they would have absolutely the same set of sounds, just some altered notes would become non-altered and vice versa. But the notation would have been much more logical and clear, without the need to calculate every time what's the actual interval between 2 notes is. (I understand that professional musicians just remember that w/o calculating, but again, the system where you have to remember it instead of just seeing is illogical. At least the visual distance between lines should differ for unequal intervals!)

In reply to by etsenberg

I like that video!
Your melody burnt into my head. I find myself humming it all the time now wandering through nature. :-)
I discovered your website and must say I'm well astonished about your german poems too. This is a very good German!
Musical language should be very easy for you to learn.
That the staff system is based on lines and spaces without reflecting the minor steps between 3-4 and 7-8 can also be seen as not-unnecessarily-overcomplicated. I bet when you sang your songs you never made a thought about where you sing major or minor intervalls. You just sang your melody up and down. And so the notes just go up and down with out any accidentals as long as your melody stays in the diatonic major or natural minor scale.
This is because the notational system derives from a singing practise AFAIK.
One could also use a more mathematical way to display notes (as in the midi-piano-rolls of DAWs), but this would make it very difficult to show all the harmonic relationships that Compositions Rules worked out.
I don't say the Staff Notation is perfect but it works very well for notating western music in a small place (paper) and to be readable also at bad light settings.
To find out or explain why we have 7 notes is so a huge topic. I have to go wandering now ;-)

In reply to by musikai

"...but most of them still use 7 notes, and I cannot understand where this number came from."

From the Greek System which stacked two 4-Tone-scales together (Tetrachords) which were built the same way:
from Top to Bottom: 1step-1step-1/2step:
a-g-f-e and e-d-c-b
The notes were found because of the physics of sound itself. The overtone scale.
Read about: Pythagoras, Overtone Scale, Guido of Arezzo
It's a fascinating topic about the physical, mathematical background of notes and sound.

In reply to by musikai

Danke, Kai :) BTW, maybe you know somebody who can sing (just for fun of course) my German songs? There are currently 2.5 of them - Landserlied, Luftwaffelied and the second half of "Take off and Fight!" (I need someone with British accent for the first half ;)) The notes can be found in my aforementioned playlist or from the music page of my website: (icons link to PDF and youtube clips).
I've read about Pythagorean scale and overtones, but I'm still not sure where 7 notes came from. They existed before Guido of Arezzo, he just made the notation (and BTW, he marked F and C line with color! I knew it was obviously logical to distinguish them someway!) 2 tetrachords of halve - tone - tone each make a beautiful symmetry until you add one more tone to complete the octave. The only mathematics which gives the number 7 I found is the following: The fifth is very important interval because it's a consonance next perfect to full octave (3/2). And in equal temperament, to make the fifth as close to 3/2 as possible it should consist of 7 of 12 equal parts of the octave. But I still don't see how from 7 halftones in the fifth 7 notes in octave follows (and equal temperament appeared only in 18th century - while of course Pythagoras or whoever else could find it intuitively). As for singing practice, I think any singer who studied solfeggio sing notes because he was taught what proper notes are, not vice versa ;)

In reply to by etsenberg

Hi etsenberg,
Yes, if I find a singer then I let you know :-) I like the Style of the Landserlied, it is a bit like the Brecht/Weill Songs.
Yes, as you found out the fifth is an important intervall also for Pythagoras :-)
I know I shouldn't do it because this is of course not the complete truth and it's better to read other sources but here is my oversimplified answer for how the 7 notes between the octaves come to shine by stacking 5ths on each other: For the ease of understanding I will start with note F:
F +5 = C
C +5 = G
G +5 = D
D +5 = A
A +5 = E
E +5 = B

Each note is octavated down to fit into 1 octave and brought into sequence.
Why did we stop here? Because the next notes will fill the gaps between the existing ones.
B +5 = F#
F# +5 = C#
C# +5 = G#
G +5 = D#
D# +5 = A#
A# +5 = E# = F
We made the circle of 5ths coming back to the initial note. Found all 12 semitones.
Of course Reality of physics always laughs as us and so in reality if you stack just perfect 5ths together the resulting F doesn't fit exactly with the intial F. This is called the Pythagorean Comma. etc.
Ages of musicians dealt with the problem to create different tunings with differently good sounding intervalls.
Today we use an equal temperament with slightly detuned smaller 5ths to be able to transpose between all 12 semitones.

You saw the nice symmetry of the Tetrachords in its early beginnings. Today you find a wonderful symmetry of Tetrachords in the major Scale which can be seen as built from 2 identical Tetrachords with the intervals:
Whole-whole-half and this again used on the (important 5th)

C D E F and starting on the 5th: G A B C

When you understand this then you can more easily understand how we find to the other Key Signatures:
So what happens when we start with the second half?
G A B C and starting on the 5th: D E F# G
There you have the reason for the needed accidental F# in the Key of G-Major.

In reply to by musikai

Yes, I've read about the circle of 5ths. In whole-tone-6-note system, where the note symbols turns from
C D E F F# G G# A A# B to
C D E E# F F# G G# A A# (remember, all symbols above E have new meaning!)
it will exist. too, turning from
F C G D A E B F# C# G# D# A# to
E# C F# D G# E A# F C# G D# A

Instead of 5 with, 7 w/o accidentals we get with-w/o-with-w/o... Looks more symmetrical. But when we start building major scales, starting from C-dur, we get
C D E E# F# G# A#
F# G# A# C D E F
D E F F# G# A# C#
G# A# C# D E F G
Thus instead of steady increasing/decreasing the number of accidentals while going around the circle we get 4-3-4-3... which is not good for key signature notation. So, if THIS is the reason to have 7 notes then we found the answer :)

In reply to by etsenberg

Haha, not understanding any of this above. :-) Why do you mention the Whole-Tone-Scale at all?
(which is
c d e f# g# a#
and not
C D E F F# G G# A A# B ) BTW.
In the concept of stackings 5th onto each other to find new notes there will never be created a Whole-Tone-Scale because the Whole-Tone-Scale doesn't have perfect 5ths anywhere. Only diminished and augmented 5ths.

But to answer your question why there are 7 notes I would say that this is just an aesthetic decision.
By concept the stacking 5ths results in the chromatic (semitone) 12-Tone-Scale.
If you only use the first 7 created notes you get a diatonic Scale.
If you only use the first 5 created you get a pentatonic Scale (with 3 whole-Tones and 2 minor 3rds: F G A C D)

In reply to by musikai

As I am saying from the beginning, we have 12 sounds (half-tones), and we can write them with 6 notes as well as with 7, only the places of accidentals will differ. That's what I did as C D E F F# G G# A A# (C C# D D# E E# F F# G G# A A# for complete 12, where the first 6 do not differ from our notes and the rest are half-tone higher, NOT our F G A. You may call them Phi Gamma Alpha to distinguish from normal F G A). But if we had this kind of notation, we would not have increasing accidentals in the circle of the 5ths, and while the scales built with the same rule of adding 7 half-tones will SOUND the same, they will LOOK, yes, less aesthetic.

In reply to by musikai

Well, as a science fiction writer I may use the alternative notation in some alternate world ;) An interesting idea BTW - I don't remember any SF book which concern alternate music, probably because most writers are not musicians.
But I could not understand the logic behind 7 notes and now I can see that building a circle of major and minor scales by adding 5ths is easier and look nicer in 7 notes system.

In reply to by Ziya Mete Demircan

Recently I saw an old science fiction movie "Close Encounters of the Third Kind". Not really worth to watch but there the scientists communicate with the aliens by playing musical sequenzes on a light tone organ.

Far more interesting is Hermann Hesses "Glasperlenspiel" which plays in a future world and features artists that play on an instrument that combines music, mathematic and philosophy and creates kind of worlds.
This book also was the introductory idea of one of Peter Neubäckers lectures on the harmonic structure of sound and interesting concepts about intervals and ideas about them in the history of important physicists and astronoms like Johannes Kepler who wrote a book about the harmonic structures of planets.
You can find some english transcripts of the lectures here:
This is the lecture about the Glasperlenspiel:

Funnily the Author Peter Neubäcker also is the programmer of Melodyne, the state of the art of polyphonic audio recognition.
This too could be used as intermediate step for mp3-to-notation but would be a quite expensive solution.

Do you still have an unanswered question? Please log in first to post your question.