Helping people with computers... one answer at a time.

Audio files contain actual sound recording while MIDI files are a type of computer program. Converting audio to midi is a nearly impossible task.

I would like to convert a .wmv file to a .mid file. All I want is the sound not the picture. I can't find any software to do this. Any ideas?

Ideas, yes. But not answers.

The problem is that there's a fundamental difference between .wmv files and .mid files - and I'm not even taking the video content into account.

They're so different that the conversion you're asking for is darned near impossible.

At least for today's computers.

I'll summarize it this way: what you're asking for is a program that can listen to a symphony and produce the sheet music that the musicians are playing from.

"Converting audio to MIDI ... is an enormous task"

To the best of my knowledge that program doesn't exist.

Now, let me explain what I mean by that statement.

Audio Recording

A .wmv file contains a recording of actual audio. Let's say it's a performance of a piano concerto; what the .wmv file contains is a recording of the sound that was produced when someone played that concerto on the piano.

Throw the video away and it's a recording of sound. Nothing more, nothing less.

In that respect, it's the same as a .mp3 file, a .wav file or any of a number of other audio formats.

A ".mid" file is very different.

MIDI

A ".mid" file, short for MIDI which stands for Musical Instrument Digital Interface is most assuredly not a recording of audio. Rather, it's a series of instructions - not unlike a computer program.

For example a .mid file for our piano concerto might have instructions like:

  • Piano: A# quarter note

  • Piano: B quarter note

  • Piano: A# half note & C half note & E half note

Note: I am nowhere near being a musician - what I just wrote probably sounds awful. Smile

In addition to my lack of musical skills, that "music" is also a gross over-simplification; MIDI is an incredibly complex and rich language for instruction what should be played when.

And therein lies the difference: MIDI is a series of instructions - notes to be played by what instrument and how.

It's not a recording of a performance; it's not sound; it's instructions.

Hence MIDI is a kind of sheet music for computers. When a computer "plays" a MIDI file, it also has to have a synthesizer of some sort - and Windows typically has a rudimentary one built in. As a result when instructed to play "Piano: A# quarter note", it uses the built in synthesizer to generate the sound of something that kind-of, sort-of sounds like a piano.

Converting One To Another

Converting a MIDI file to sound is relatively easy: you run it through a midi player which drives either a synthesizer, or some computer-enabled instruments that then play the instructions contained in the file. If you want an audio recording file like a .mp3, you simply record the performance in some way.

Converting audio to MIDI, on the other hand, is an enormous task. It really is like listening to our piano concerto and writing down as it goes the extremely detailed instructions that would have resulted in that performance.

And that's just a piano concerto - one instrument. Imagine trying to write down the sheet music from simply listening to a symphony. Or a rock song.

It's just not something that can be done today.

Article C4484 - October 9, 2010

Leo Leo A. Notenboom has been playing with computers since he was required to take a programming class in 1976. An 18 year career as a programmer at Microsoft soon followed. After "retiring" in 2001, Leo started Ask Leo! in 2003 as a place for answers to common computer and technical questions. More about Leo.

Not what you needed?

Recent Comments
15 Comments

Sounds like Leo is saying that midi is to music what Postscript is to printing. A language
to describe what should be created.

Posted by: Fuzzy at October 12, 2010 6:56 PM

Not so, all wrong I'm afraid! The question appears to be clear, how to convert an audio file (.wmv in this case), to midi data or a .mid file?

Its a process nearly all midi users (apart from expert sequencers perhaps), would like to have access to, since it enables the generation of a midi sequence from an audio file such as an mp3 or wav file (say from a commercially available CD).

The problem is that the audio file is a polyphonic and multi-instrument source and the best that the most of the software that is available (to convert audio signals into midi data), can do, is to produce a midi data sequence from a single sound or instrument source.

The problem is (and always has been), that there are problems with pitch recognition and tracking of fast passages of notes, which is compounded by the fact that what is required to make this conversion effective and thus produce a useable midi file, is that the software must simultaneously recognise the pitches and note sequences of many instruments and produce an accurate sequence of midi data for each of these instruments, preferably in separate tracks (or more correctly, separate midi channels).

There are some examples of software that can convert an audio source, such as a microphone's output or electric guitar etc, into midi data sequences, but (for the reasons given above), they cannot (yet), consistently produce a useable midi sequence from an audio source which includes multi-instruments.

The software is getting better (I think "MagicEar" is one of the leaders in this field), it manages to produce multi channel, midi sequences from a multi-instrument audio source, but the process is difficult and frequently produces inconsistent or wrong results.

Its remarkable that it can be achieved at all, since the software has to guess at which instrument is which (ie separate the instruments out on the basis of pitch range, unique characteristics etc), and then has to simultaneously identify the pitch and duration of each note for each instrument and generate the appropriate, separate midi channel containing the instructions for the midi sound source to reproduce an approximation of the original audio sounds.

Long winded response I know, but midi is one of my passions. The "Intelliscore" software is different altogether, what it does is recognise notes etc (using Optical Character Recognition processes), from a musical score and generate the appropriate midi data to enable a midi sequencer to play back the score, it also has problems with multi tracks, recognition of the charaters and correctly generating the appropriate pitches and note divisions etc, but its easier than audio to midi, however you have to have the complete, accurate musical score of the audio file (song?), you wish to produce a midi sequence (file), for.

Good luck finding the software, I've been looking and trying for years, please let me know if you find something that works consistently and accurately.

PS: you CAN convert midi to audio, any midi sound source (or tone generator if you like), can do it (including, these days), the most basic computer sound cards. I think "RJ" might have stated it the wrong way around, in which case its difficult but not impossible.

There is software available that can generate a music score from a midi file, but not so far as I am aware and, as Leo says, from an audio source.

Posted by: Richard at October 12, 2010 7:54 PM

Update to my previous post: I did not recall the names of the software programs correctly and was thinking of "Midi Notate" when I commented on "Intelliscore" saying that it was just a score recognition program. Similarly, I was thinking of "Digital Ear" when I said "Magic Ear".

Intelliscore is now in its 8th edition and if you try it out to convert even a simple (meaning few instruments, say a "rock quartet" - one of the ensembles Intelliscore suggests for multi-instrument conversion), audio track from a CD or an mp3 file you will find that the results are unuseable as a music track or backing sequence.

The timing is all over the place, the pitch is similarly inaccurate, the instruments are merged together and the relative volume of the various instruments appears to vary with the "beat" of the music, some drop out altogether.

Given the variables involved and the problems in getting a machine (computer + program and microphone), to be able to listen, identify and discern the difference between the individual intstruments in an "ensemble" the way that the human ear can, appears to be impossible at the present time.

Posted by: richard at October 13, 2010 12:03 AM

Midi's are so easy to find. just find the music in the midi format to start with. Then it is easy to convert it to another format.

Posted by: Glen at October 25, 2010 3:16 PM

melodyne will convert audio to midi... for example, if i play a guitar solo, then it can convert the notes to their midi counterparts (does a pretty decent job)... the basic version of melodyne works only with monophonic sounds, but there is another version that will even deconstruct polyphonic stuff like chords and all that mess... yonzers

Posted by: andrew at December 19, 2010 11:32 AM
Post a comment on "How do I convert an audio file to a midi file?":





Remember Me?

(You may use HTML tags for style)

Before commenting, please...

  • READ THE ARTICLE. A comment that shows you didn't will be deleted and ignored.

  • Comment only on the article. Use the search box at the top of the page if you have a question about something else.

  • NO PERSONAL INFORMATION in the comment. No email addresses. No phone numbers. No physical addresses.

  • Anything that looks the least bit like spam will be deleted. Links to unrelated sites or links that appear to be primarily promotional will be deleted, or the comment will be deleted.

  • Don't ask me to recover lost passwords or hacked accounts. I can't. Those comments will be deleted.

  • I can't respond to every comment. And I can't vouch for the accuracy of others who do.

Please wait. Your comment is being processed ...