Mechanical Music Digest  Archives
You Are Not Logged In Login/Get New Account
Please Log In. Accounts are free!
Logged In users are granted additional features including a more current version of the Archives and a simplified process for submitting articles.
Home Archives Calendar Gallery Store Links Info
MMD > Archives > August 2002 > 2002.08.28 > 03Prev  Next


Audio-to-MIDI Conversion
By Julie Porter

[ Bernt Damm wrote: ]

> As an electronics engineer, I would say that this should be possible,
> although expensive.  It should be possible to construct 88 filters
> that analyze the audio signal from microphone and give an output if
> that particular frequency is present.  To make that into MIDI would
> or sheet music or piano rolls would just be a software exercise.
> The modern way to do this is probably a fast DSP (digital signal
> processor) to simulate the filters.

I would like to point out that there has been a newsgroup on Usenet
which is accessible from Google groups that can be found as
<news//alt.music.midi>.  This group published a FAQ.  The group has
been in existence for many years.  (I have been on Usenet since it was
part of the Arpanet, for over 25 years.)  AAM has been in existence at
least 10 years.

What you describe is a set of mathematics of a class called Fourier
transforms.  A DSP is based on a class called the Fast Fourier
Transform (FFT).

Recently someone has been spamming a lot of the Internet with a "Gold
Brick" that is a convincing demo of this much demanded program.

The FAQ at alt.music.midi is trying to warn against this snake oil;
however, the masses so want this to be true that there are thousands of
people who will not listen to Cassandra.

1) The program only works with a few seconds of audio.

2) The authors will not let it be tested in a university setting with
sound sample of the investigators choice.  In other words, it fails the
"double blind test" of researchers who really are working on this
"software problem."  The main test is to take the waveform of a
didgeridoo and replicate the fingering note events.

3) The program uses the MIDI "pitch bend" controller and continuous
notes to "match" the playback.  This makes the resulting MIDI totally
un-editable and useless for the detection of individual events.

A recent poster to AMM called it looking like "spasmodic MIDI" or "MIDI
on Drugs", and totally useless for score generation.  For piano or band
organ playback the file is effectively blank.  The resulting piano would
turn on a set of notes, then burn out the solenoids as there is no way
to "pitch bend" any solenoid driver, so these MIDI events are ignored.

4) The program is only available cash in advance!

5) The FFT/DCT/DSP is a one-way lossy compression.  Case in point: When
text editors started becoming popular, a newspaper was showing it off
live on the local TV program.  "Look," he said, "I can change all the
'e' characters on tomorrows front page in the file to the 't'
character.  Now watch me change them back."  (!!!)

> Problems I can see with this are the harmonics in a piano and also
> other instruments, voice and singing.  Nevertheless, I am convinced
> it is possible.

So are about a million other folks, including a few hundred spam artists.
"The proof is in the pudding," or, in this case, the proof is on the
music sheet.

All I can say is "Caveat Emptor".  If it sounds too good to be true,
it probably is.

Julie Porter

 [ Julie is an experienced programmer now working at a firm that makes
 [ MIDI devices.  Before that she worked at Apple Computer.  -- Robbie


(Message sent Wed 28 Aug 2002, 19:39:32 GMT, from time zone GMT-0700.)

Key Words in Subject:  Audio-to-MIDI, Conversion

Home    Archives    Calendar    Gallery    Store    Links    Info   


Enter text below to search the MMD Website with Google



CONTACT FORM: Click HERE to write to the editor, or to post a message about Mechanical Musical Instruments to the MMD

Unless otherwise noted, all opinions are those of the individual authors and may not represent those of the editors. Compilation copyright 1995-2024 by Jody Kravitz.

Please read our Republication Policy before copying information from or creating links to this web site.

Click HERE to contact the webmaster regarding problems with the website.

Please support publication of the MMD by donating online

Please Support Publication of the MMD with your Generous Donation

Pay via PayPal

No PayPal account required

                                     
Translate This Page