There is a MP3 to Mid Converter to me use on a Remix?

[Hidan]

The deal is, I want to make a remix of a song, but this song is only available in .mp3 file, and I can't sample a mp3 file like I would can with a mid file.

[Rozovian]

You're better off transcribing the parts you need by ear. Any audio->midi tool will get distracted by all the notes in there, and their different timbre, and all the effects, and it can't really tell instruments apart anyway like the human mind can.

There are audio->midi tools, but they tend to work best with monophonic (as in only one note at a time) materials, and even then they get stuff wrong. The exception would be software like Melodyne, but I'm not sure how well it tells instruments apart and what quality of audio it needs (a 200kbps mp3 might not be accurate enough for it to really discern the notes).

[Hidan]

Ok, thanks anyway.

[Nabeel Ansari]

Considering mp3 is compressed audio, I don't think there's any such software that can extract MIDI data (which is instructions or "computer sheet music", not audio) from it.

I don't think it'll be possible for quite some time.

[Hidan]

Considering mp3 is compressed audio, I don't think there's any such software that can extract MIDI data (which is instructions or "computer sheet music", not audio) from it.

I don't think it'll be possible for quite some time.

I found this program called Amazing MiDi. It converts Wav to Mid. If I convert MP3 to Wav and then convert the Wav to Mid, it could work?

The site of this program: http://www.pluto.dti.ne.jp/~araki/amazingmidi/

Edit: Nevermind... It didn't work. =P

[Rozovian]

The reason it's not working is probably because the audio is too complex. In a single note, there's loads of frequencies at different amplitudes (volume) that behave differently as the note decays. In a song, you've got loads of notes from loads of instruments, and they're processed with various effects. It's like a few dozen semi-transparent pictures on top of each other. It's difficult to tell what belongs to which picture.

It doesn't help that mp3 is a compressed audio format, basically a low-quality snapshot of the real song. Can't give it back the detail by converting it back to wav. For complex audio, you kind'a need that detail if a program's supposed to understand the difference between the fundamental frequency of a note and a frequency that's an overtone, resonance, or otherwise not a fundamental.

You don't have to understand this all, just know that you'll get better results from training your own ears and practicing transcribing that you're gonna get with programs. Most of the time. After some practice, it's not nearly as difficult as it is when you start out. If you can hum it, you can write it.

[Hidan]

The reason it's not working is probably because the audio is too complex. In a single note, there's loads of frequencies at different amplitudes (volume) that behave differently as the note decays. In a song, you've got loads of notes from loads of instruments, and they're processed with various effects. It's like a few dozen semi-transparent pictures on top of each other. It's difficult to tell what belongs to which picture.

It doesn't help that mp3 is a compressed audio format, basically a low-quality snapshot of the real song. Can't give it back the detail by converting it back to wav. For complex audio, you kind'a need that detail if a program's supposed to understand the difference between the fundamental frequency of a note and a frequency that's an overtone, resonance, or otherwise not a fundamental.

You don't have to understand this all, just know that you'll get better results from training your own ears and practicing transcribing that you're gonna get with programs. Most of the time. After some practice, it's not nearly as difficult as it is when you start out. If you can hum it, you can write it.

I see. Kind of a difficult job isn't it?

But that's ok, i'll try to focus on the parts of the music and manually put each instrument segment on their piano rolls.

[Gario]

There should really be a sticky on this, or people should search OCR, since I've seen this question asked multiple times before.

The short answer is: No, there isn't. The long answer is: No, because you are requesting a computer to translate a compound sound wave into a recognizable program called 'Midi', in which the computer is programed to output sounds based on simple 128 bit info. Waveforms are incredibly complex, so much so that in many cases when multiple things are playing it is simply not possible to read what notes are playing, because the combinations of waves create patterns that look similar whether it's an A5, F#3, etc., due to timbre variance. The variables that influence the waves can make two different pitches look like similar waves, so a computer will never be able to disambiguate such waveforms. It hasn't happened, and I suspect that unless we can devise another way for computers to read uncompressed (or compressed) files that program will never exist.

So you'll need to do what everyone else does and dictate the music; that is, listen to it on your own and write it down based on what you hear. Give it a little practice and you'll get the hang of it just fine.

[SonicThHedgog]

i had one that was excellent and it converted mp3 to midi in tracks but was bunching parts (and drums) on some songs turing it to a spaghetti midi that did not look nice some times but it did help some time,

to you question, Yes and No, No because there not so stable *YET* Yes because some work (OK) but as tech improvs over time, maybe we will have a legit extremly stable Mp3/wav to midi. but for now compose the midi by ear.

[Nabeel Ansari]

i had one that was excellent

...

No because there not so stable

...lolwut?