back to list

Audio to MIDI conversion

🔗Yahya Abdal-Aziz <yahya@melbpc.org.au>

4/14/2005 7:20:12 PM

Hi,

I'm changing the subject line, as this is a tangent to
the original:
Improvisation on pitches from a recording of a song.

Automated conversion of audio to notation or MIDI has
long been an interest of mine; as has bird-song, since we
have so many beautiful native and introduced songbirds
here. I believe Olivier Messiaen based much of his musical
theory and practice on birdsong (more particularly in the
realm of additive rhythms than in pitch and scale).

Anyway this recent discussion, started by Robert Walker
prompted me to do a little more research.

Ozan wrote:
____________________________________________
Solo Explorer 1.0 from www.recognisoft.com has a modest
pitch to midi conversion feature for monophonic wave files.

----- Original Message -----
From: "Graham Breed"
....
>
> Robert Walker wrote:
>
> > BTW one idea I had for analysing bird song
> > is to try and fit a best fit sine wave to it.
>
> Have you looked at Csound? Chapter 21 of "The Csound Book" describes
> how to make a pitch to MIDI convertor. I haven't tried it out, but it
> looks like a lot of thought has gone into the pitch tracking utilities.
> I think the source code is LGPL, so you can use it, but with
restrictions.
>
> It's a few weeks since I read it, but I think they do best fits for the
> full harmonic series.
>
_______________________________________________________________________

[Yahya ] I've previously used Transcribe 0.9 - monophonic,
and briefly, TSWav2Midi (from Russia I think), which
claims to successfully extract useful pitch info from
polyphonic music. But I can't find it now, nor a link to it.

I've just done a KartOO search and have some results for you:

Monophonic - MuseBook Wav2Midi is at:
http://www.musebook.com/?page=mbwav2midi

Also, Digital Ear Real-Time, at:
http://www.audiofilesland.com/company/epinoisis-software/digital-ear-real-ti
me.html
- claims "Unlike conventional so-called "Pitch-to-MIDI"
converters, Digital Ear will send high-resolution pitch events
closely matching those of your original sound. Any vibrato,
tremolo, pitch-bend, or portamento effects of your recorded
sound will be faithfully converted and reproduced into any
voice of your synthesizer." This may obscure the scales you're
looking for, and appears to be mono anyway.

Polyphonic - Intelliscore Polyphonic WAV to MIDI Converter at:
http://www.audiofilesland.com/company/innovative-music-systems-inc/intellisc
ore-polyphonic-wav-to-midi-converter.html

Widisoft, at: http://www.widisoft.com/?ref=adw_eng looks
interesting. It claims to produce a notated score from an input
MP3 or WAV file.

Wavetomidi is for Windows and Linux, unregistered version has
full features, & from the sample files included, appears to be
polyphonic. It's at:
http://harmonicbits.com/en/soundpr/w2m.html

I've decided to download these last two for a try out, and will
report back to the list on them.

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.9.9 - Release Date: 13/4/05

🔗Carl Lumma <ekin@lumma.org>

4/14/2005 10:54:57 PM

Hi Yahya,

I've used or tried many of these in the past. Recognisoft Solo
Explorer is my last hope for a microtonal-friendly one (it claims
it's good for "intonation research" but I can't get the demo to
work at the moment).

Transcribe! is a great piece of software, and I have used it to
test the tuning of Barbershop chords, but the tuning readout is
on mouseover only. I requested a better tuning readout some time
ago and things did improve with version 7, but not enough to make
much of a difference.

Digital Ear...

http://digital-ear.com

...is another great piece of software, but it too obfuscates the
cents readout.

Widisoft has a fantastic interface, but its transcriptions snap
to 12-tET no matter how I configure the recognition (even for
monophonic samples).

-Carl

>[Yahya ] I've previously used Transcribe 0.9 - monophonic,
>and briefly, TSWav2Midi (from Russia I think), which
>claims to successfully extract useful pitch info from
>polyphonic music. But I can't find it now, nor a link to it.
>
>I've just done a KartOO search and have some results for you:
>
>Monophonic - MuseBook Wav2Midi is at:
>http://www.musebook.com/?page=mbwav2midi
>
>Also, Digital Ear Real-Time, at:
>http://www.audiofilesland.com/company/epinoisis-software/
>digital-ear-real-time.html
>- claims "Unlike conventional so-called "Pitch-to-MIDI"
>converters, Digital Ear will send high-resolution pitch events
>closely matching those of your original sound. Any vibrato,
>tremolo, pitch-bend, or portamento effects of your recorded
>sound will be faithfully converted and reproduced into any
>voice of your synthesizer." This may obscure the scales you're
>looking for, and appears to be mono anyway.
>
>Polyphonic - Intelliscore Polyphonic WAV to MIDI Converter at:
>http://www.audiofilesland.com/company/innovative-music-systems-inc/
>intelliscore-polyphonic-wav-to-midi-converter.html
>
>Widisoft, at: http://www.widisoft.com/?ref=adw_eng looks
>interesting. It claims to produce a notated score from an input
>MP3 or WAV file.
>
>Wavetomidi is for Windows and Linux, unregistered version has
>full features, & from the sample files included, appears to be
>polyphonic. It's at:
>http://harmonicbits.com/en/soundpr/w2m.html
>
>I've decided to download these last two for a try out, and will
>report back to the list on them.
>
>Regards,
>Yahya

🔗Yahya Abdal-Aziz <yahya@melbpc.org.au>

4/15/2005 8:17:33 PM

Carl,

Thanks for the feedback.

Have you yet tried MuseBook Wav2Midi,
Intelliscore Polyphonic WAV to MIDI Converter,
or Wavetomidi?

I'm still quite hopeful of this last one.

Regards,
Yahya

-----Original Message-----
________________________________________________________________________
Date: Thu, 14 Apr 2005 22:54:57 -0700
From: Carl Lumma <ekin@...>

Hi Yahya,

I've used or tried many of these in the past. Recognisoft Solo
Explorer is my last hope for a microtonal-friendly one (it claims
it's good for "intonation research" but I can't get the demo to
work at the moment).

Transcribe! is a great piece of software, and I have used it to
test the tuning of Barbershop chords, but the tuning readout is
on mouseover only. I requested a better tuning readout some time
ago and things did improve with version 7, but not enough to make
much of a difference.

Digital Ear...

http://digital-ear.com

...is another great piece of software, but it too obfuscates the
cents readout.

Widisoft has a fantastic interface, but its transcriptions snap
to 12-tET no matter how I configure the recognition (even for
monophonic samples).

-Carl

>[Yahya ] I've previously used Transcribe 0.9 - monophonic,
>and briefly, TSWav2Midi (from Russia I think), which
>claims to successfully extract useful pitch info from
>polyphonic music. But I can't find it now, nor a link to it.
>
>I've just done a KartOO search and have some results for you:
>
>Monophonic - MuseBook Wav2Midi is at:
>http://www.musebook.com/?page=mbwav2midi
>
>Also, Digital Ear Real-Time, at:
>http://www.audiofilesland.com/company/epinoisis-software/
>digital-ear-real-time.html
>- claims "Unlike conventional so-called "Pitch-to-MIDI"
>converters, Digital Ear will send high-resolution pitch events
>closely matching those of your original sound. Any vibrato,
>tremolo, pitch-bend, or portamento effects of your recorded
>sound will be faithfully converted and reproduced into any
>voice of your synthesizer." This may obscure the scales you're
>looking for, and appears to be mono anyway.
>
>Polyphonic - Intelliscore Polyphonic WAV to MIDI Converter at:
>http://www.audiofilesland.com/company/innovative-music-systems-inc/
>intelliscore-polyphonic-wav-to-midi-converter.html
>
>Widisoft, at: http://www.widisoft.com/?ref=adw_eng looks
>interesting. It claims to produce a notated score from an input
>MP3 or WAV file.
>
>Wavetomidi is for Windows and Linux, unregistered version has
>full features, & from the sample files included, appears to be
>polyphonic. It's at:
>http://harmonicbits.com/en/soundpr/w2m.html
>
>I've decided to download these last two for a try out, and will
>report back to the list on them.
>
>Regards,
>Yahya

________________________________________________________________________

--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.9.13 - Release Date: 16/4/05

🔗Carl Lumma <ekin@lumma.org>

4/15/2005 8:59:26 PM

>Carl,
>
>Thanks for the feedback.
>
>Have you yet tried MuseBook Wav2Midi,
>Intelliscore Polyphonic WAV to MIDI Converter,
>or Wavetomidi?
>
>I'm still quite hopeful of this last one.
>
>Regards,
>Yahya

I tried Intelliscore back in 2000, and predicted that
by now it would be working at human levels. Too bad
I was so wrong.

I tried it again in 2002 I think, but it looks like
some improvements have been made since then.

I tried Wavetomidi just yesterday, I think it was,
and I wasn't impressed. Definitely amateur night.

I didn't try MuseBook, but maybe I should...

If anyone is able to e-mail recognisoft or to get
their Solo Explorer demo to work, I'd love to hear
about it.

-Carl

🔗Yahya Abdal-Aziz <yahya@melbpc.org.au>

4/17/2005 5:08:38 PM

Hi Carl and others,

Report on my trials (and tribulations!) as promised:

Wavetomidi 2.2
------------------

This program is VERY disappointing, as it can't manage
to recognise half the pitches in some simple polyphony,
no matter how I tweak the settings.

I started with the Froberger music samples from the
Huygens-Fokker website, but realising these might be
a bit sophisticated in terms of envelope, organ
acoustics and concurrent polyphony, have tried instead
to have Wavetomidi analyze a piece of my own, (Longing,
for viola and guitar), which is fairly transparent writing.

Results? Woeful!

Widi 3.21
-----------

Carl, I agree with you on the interface - it's VERY good
indeed.

I tried it on "Longing" and it had no trouble getting most
of the notes right with the default settings - certainly
enough to begin a transcription. Much, much better than
Wavetomidi.

There are options galore, but they're well organised, so
that you can get results very quickly using the Wizard that
shows by default at startup. Yet there seems to be no
recognition of anything other than 12-tET.

Carl wrote that:
"Widisoft has a fantastic interface, but its transcriptions
snap to 12-tET no matter how I configure the recognition
(even for monophonic samples)."

According to the readme.txt file for Widi 3.21:
"Please, inform us about bugs and your ideas on how to
improve this Program. (Regardless of whether you are a
registered user or not). Language corrections to the
documentation are also welcome. Write to us if you want
to participate in this project.

You can find the recent versions at:
www.widisoft.com

Our e-mails are:
General Info info@widisoft.com
Customer Support support@widisoft.com

Music Recognition Team 12.01.2005"

It may be that they would be open to, and interested in,
a request for microtonal support. Have you tried
contacting them, Carl?

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.9.15 - Release Date: 16/4/05

🔗Carl Lumma <ekin@lumma.org>

4/17/2005 5:47:19 PM

>Wavetomidi 2.2
>------------------
>
>This program is VERY disappointing, as it can't manage
>to recognise half the pitches in some simple polyphony,
>no matter how I tweak the settings.
>
>I started with the Froberger music samples from the
>Huygens-Fokker website, but realising these might be
>a bit sophisticated in terms of envelope, organ
>acoustics and concurrent polyphony, have tried instead
>to have Wavetomidi analyze a piece of my own, (Longing,
>for viola and guitar), which is fairly transparent writing.
>
>Results? Woeful!

Same here.

>According to the readme.txt file for Widi 3.21:
>"Please, inform us about bugs and your ideas on how to
>improve this Program.
//
>It may be that they would be open to, and interested in,
>a request for microtonal support. Have you tried
>contacting them, Carl?

I haven't. May I suggest you have at it?

-Carl

🔗monz <monz@tonalsoft.com>

4/18/2005 12:52:09 AM

--- In tuning@yahoogroups.com, "Yahya Abdal-Aziz" <yahya@m...> wrote:
>
> Hi Carl and others,
>
> Report on my trials (and tribulations!) as promised:
>
> <snip>
>
> Widi 3.21
> -----------
>
> Carl, I agree with you on the interface - it's VERY good
> indeed.
>
> I tried it on "Longing" and it had no trouble getting most
> of the notes right with the default settings - certainly
> enough to begin a transcription. Much, much better than
> Wavetomidi.
>
> There are options galore, but they're well organised, so
> that you can get results very quickly using the Wizard that
> shows by default at startup. Yet there seems to be no
> recognition of anything other than 12-tET.
>
> Carl wrote that:
> "Widisoft has a fantastic interface, but its transcriptions
> snap to 12-tET no matter how I configure the recognition
> (even for monophonic samples)."
>
> According to the readme.txt file for Widi 3.21:
> "Please, inform us about bugs and your ideas on how to
> improve this Program. (Regardless of whether you are a
> registered user or not). Language corrections to the
> documentation are also welcome. Write to us if you want
> to participate in this project.
>
> You can find the recent versions at:
> www.widisoft.com
>
> Our e-mails are:
> General Info info@w...
> Customer Support support@w...
>
> Music Recognition Team 12.01.
2005"
>
> It may be that they would be open to, and interested in,
> a request for microtonal support. Have you tried
> contacting them, Carl?

several years ago, i too was greatly impressed by the results
of converting .wav files to MIDI with WIDI. listen to this:

/tuning/files/monz/gdtimes2.mid

if you're familiar with the music, the Opening Theme from the
TV show "Good Times", you'll recognize that my "orchestra of
ocarinas" playing the WIDI output comes amazingly close to
actual human voices singing actual English words. yes, there
are weird artifacts in the form of many very high notes which
aren't really part of the .wav file, and are simply the result
of WIDI doing its FFT analysis on the .wav file. but i swear
that it's almost possible to make out the words, and all this
is, is a bunch of notes in a MIDI file played in 12-edo.
you can read the words at the bottom left of this page:

http://valdefierro.com/times05.html

i wrote to the folks at WIDI then, encouraging them to make
the WIDI output contain pitch-bends which would tune the
higher partials more accurately. their response was enthusiastic,
but non-committal ... and apparently they never did take up
the suggestion.

-monz

🔗Yahya Abdal-Aziz <yahya@melbpc.org.au>

4/19/2005 8:29:33 AM

Carl,

You wrote:

> >According to the readme.txt file for Widi 3.21:
> >"Please, inform us about bugs and your ideas on how to
> >improve this Program.
//
> >It may be that they would be open to, and interested in,
> >a request for microtonal support. Have you tried
> >contacting them, Carl?

> I haven't. May I suggest you have at it?

With respect, I think you'd be better placed to advise them
- should they choose to ask! - on which of the many possible
tuning systems would be of most use: n-EDO? Meantone?
Almost certainly, you also know a great many more microtonal
works, and workers, than I do.

I'm not shy (as I think you know) but if we're looking for a
result, we want to make the best possible case.

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.9.15 - Release Date: 16/4/05

🔗Yahya Abdal-Aziz <yahya@melbpc.org.au>

4/19/2005 8:29:45 AM

Hi all,

Reason I couldn't find TSWavtoMIDI is because I
had misremembered the name. The correct name is
TS-AudioToMIDI. It was at version 3.30, when I
downloaded the 30-day trial version on 1 January
2005. Yes, it's expired! :-) I'll have another go.

According to the Readme.txt file, it claims to offer -

o polyphonic recognition;
o separate display of the spectrum;
o user specification of the first four partials
(is that enough to be useful for any real
instrument?);
o several different recognition algorithms;
o filtering of small note durations;
and
o beat detection.

But it talks of using "any external MIDI editor", so
don't expect it to be an all-in-one solution.

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.9.15 - Release Date: 16/4/05

🔗Yahya Abdal-Aziz <yahya@melbpc.org.au>

4/19/2005 8:29:42 AM

monz,

You wrote:
________________________________________________________________________

several years ago, i too was greatly impressed by the results
of converting .wav files to MIDI with WIDI. listen to this:

/tuning/files/monz/gdtimes2.mid

if you're familiar with the music, the Opening Theme from the
TV show "Good Times", you'll recognize that my "orchestra of
ocarinas" playing the WIDI output comes amazingly close to
actual human voices singing actual English words. yes, there
are weird artifacts in the form of many very high notes which
aren't really part of the .wav file, and are simply the result
of WIDI doing its FFT analysis on the .wav file. but i swear
that it's almost possible to make out the words, and all this
is, is a bunch of notes in a MIDI file played in 12-edo.
you can read the words at the bottom left of this page:

http://valdefierro.com/times05.html

i wrote to the folks at WIDI then, encouraging them to make
the WIDI output contain pitch-bends which would tune the
higher partials more accurately. their response was enthusiastic,
but non-committal ... and apparently they never did take up
the suggestion.
________________________________________________________________________

Don't know the "Good Times" music, but will check it out
when I go online. (Composing replies offline saves scarce
connect time at 28.8kbps max ... :-( )

"Enthusiatic, but non-committal", eh?

That's right up there with "all care and no responsibility"
or "fair-weather friends" ...

Perhaps they don't (yet) see any commercial value in doing so?
Maybe we need to do some marketing, make microtonal music
_fashionable_, with an astute and sexy marketer as its front
gal or guy ... I know, let's ask Madonna to go microtonal - she'll
try anything - on camera even! :-)

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.9.15 - Release Date: 16/4/05

🔗monz <monz@tonalsoft.com>

4/19/2005 10:44:02 AM

hi Yahya,

> Don't know the "Good Times" music, but will check it out
> when I go online. (Composing replies offline saves scarce
> connect time at 28.8kbps max ... :-( )

i think you too will be amazed at how well the WIDI output
approximates the original .wav file, which you can probably
still find somewhere on the internet.

> "Enthusiatic, but non-committal", eh?
>
> That's right up there with "all care and no responsibility"
> or "fair-weather friends" ...
>
> Perhaps they don't (yet) see any commercial value in doing so?
> Maybe we need to do some marketing, make microtonal music
> _fashionable_, with an astute and sexy marketer as its front
> gal or guy ... I know, let's ask Madonna to go microtonal - she'll
> try anything - on camera even! :-)

well, my point about including pitch-bends was so that the
WIDI output, rather than using only 12-edo MIDI notes, would
be tweaked so that those high MIDI notes would give a better
approximation to the upper partials being plopped out of
WIDI's FFT analysis. the idea was that the WIDI output
could be made to sound even more like the original .wav input
if it used pitch-bends.

i have to say that i've downloaded later versions of WIDI
since then and have been disappointed to find that the
output was not as good as the older version ... but i also
must admit that i didn't spend a lot of time playing
around with the settings.

-monz

🔗Carl Lumma <ekin@lumma.org>

4/19/2005 10:28:04 AM

>Carl,
>
>You wrote:
>
>> >According to the readme.txt file for Widi 3.21:
>> >"Please, inform us about bugs and your ideas on how to
>> >improve this Program.
>//
>> >It may be that they would be open to, and interested in,
>> >a request for microtonal support. Have you tried
>> >contacting them, Carl?
>
>> I haven't. May I suggest you have at it?
>
>With respect, I think you'd be better placed to advise them
>- should they choose to ask!

They *did* ask (above).

>- on which of the many possible tuning systems would be of
>most use: n-EDO? Meantone?

Usually all EDO can be implemented with the same code.
Linear temperaments such as meantone and JI are harder
to explain, but since they can be approximated to arbitrary
accuracy with EDOs, EDOs alone suffice.

>Almost certainly, you also know a great many more microtonal
>works, and workers, than I do.
>
>I'm not shy (as I think you know) but if we're looking for a
>result, we want to make the best possible case.

I'm sure you'd do fine. I'm not a registered Widi user -- are
you?

-Carl

🔗Carl Lumma <ekin@lumma.org>

4/19/2005 11:24:43 AM

>Hi all,
>
>Reason I couldn't find TSWavtoMIDI is because I
>had misremembered the name. The correct name is
>TS-AudioToMIDI. It was at version 3.30, when I
>downloaded the 30-day trial version on 1 January
>2005. Yes, it's expired! :-) I'll have another go.
>
>According to the Readme.txt file, it claims to offer -
>
>o polyphonic recognition;
>o separate display of the spectrum;
>o user specification of the first four partials
> (is that enough to be useful for any real
> instrument?);
>o several different recognition algorithms;
>o filtering of small note durations;
>and
>o beat detection.
>
>But it talks of using "any external MIDI editor", so
>don't expect it to be an all-in-one solution.
>
>Regards,
>Yahya

I've had that installed for months, but I just now
activated the trial. 30 days and ticking. I'll try
to play with it tonight. ... Ok, maybe later this
week.

-Carl

🔗Carl Lumma <ekin@lumma.org>

4/20/2005 12:16:02 AM

> several years ago, i too was greatly impressed by the results
> of converting .wav files to MIDI with WIDI. listen to this:
>
> /tuning/files/monz/gdtimes2.mid
>
> if you're familiar with the music, the Opening Theme from the
> TV show "Good Times", you'll recognize that my "orchestra of
> ocarinas" playing the WIDI output comes amazingly close to
> actual human voices singing actual English words.

Wow; neat. It's working like an 88-band vocoder with center
frequencies (or rather floor frequencies) in 12-tET.

-Carl

🔗Carl Lumma <ekin@lumma.org>

4/21/2005 9:05:16 PM

http://tinyurl.com/7pfpo?__ZenphStudios

-C.

🔗Ozan Yarman <ozanyarman@superonline.com>

4/21/2005 9:46:29 PM

Fascinating.
----- Original Message -----
From: Carl Lumma
To: tuning@yahoogroups.com
Sent: 22 Nisan 2005 Cuma 7:05
Subject: Re: [tuning] Re: Audio to MIDI conversion

http://tinyurl.com/7pfpo?__ZenphStudios

-C.