20 Nov
2008
20 Nov
'08
10:09 a.m.
Hi all,
I'm currently working on Latin patterns for the Drummer's Free Art and
consequently with the percussions the project becomes sharper.
I wrote each instrument in its own track, with some dynamics events, as
Lilypond allows that.
My problem is Lilypond creates for each dynamic event a midi CC7 event
instead of a velocity event.
Is it possible to transform these CC into Velocity events?
I thought about mididings. I'm able to run and appreciate it at home but
i'm not knowledgable in python if i want to write my own script and
redirect the new midi file.
Is there any light solution?
PS. I join an example with my post. If you open it in a sequencer and
look at the bassdrum you'll figure out what i say.
Thanks in advance.
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
20 Nov
20 Nov
4:50 p.m.
On Thursday 20 November 2008 15:14:00 Philippe Hezaine wrote:
My problem is Lilypond creates for each dynamic event
a midi CC7 event
instead of a velocity event.
Is it possible to transform these CC into Velocity events?
I suppose by "velocity events" you mean the velocity values of subsequent
note-on events?
I thought about mididings. I'm able to run and
appreciate it at home but
i'm not knowledgable in python if i want to write my own script and
redirect the new midi file.
I'm not sure mididings is the right tool for the job. The Python code to
transform the events would be quite simple, but mididings doesn't support
reading/writing MIDI files (yet?). You could of course use a separate MIDI
player/recorder, and route the events through mididings, but you'd lose all
meta information (tempo, time signature, etc.) from the original file.
Having said that, I don't really know a better solution.
midish (http://caoua.org/midish/) can import/export MIDI files, but I'm not
sure how/if it can do the transformation you want.
PS. I join an example with my post. If you open it in
a sequencer and
look at the bassdrum you'll figure out what i say.
As far as I can see, all tracks in your MIDI file are on the same channel,
so CC #7 messages on one track will affect the other tracks as well. You
should probably assign a different channel to each track.
Cheers,
Dominic
6:44 p.m.
Dominic Sacré a écrit :
Yes. I was sucessfull to loop a midi file with Jack and Rosegarden. The
outputs was routed through mididings and went back to Rg for a new
record on a new track. But it was not for this purpose and i didn't look
for the meta information.
But for now, if i have a copy about these informations may be i can try
again this experience? It isn't a piece of music, only a succession of
drums patterns. I can organize them to avoid different time signature...
No. Like i have said these are drums patterns written by Lilypond. After
the famous transformation i could join all the tracks in one if i want.
On Thursday 20 November 2008 15:14:00 Philippe Hezaine
wrote:
Yes. Sorry.
My problem is Lilypond creates for each dynamic
event a midi CC7 event
instead of a velocity event.
Is it possible to transform these CC into Velocity events?
I suppose by "velocity events" you mean the velocity values of subsequent
note-on events?
I thought
about mididings. I'm able to run and appreciate it at home but
i'm not knowledgable in python if i want to write my own script and
redirect the new midi file.
I'm not sure mididings is the right tool for the job. The Python code to
transform the events would be quite simple, but mididings doesn't support
reading/writing MIDI files (yet?). You could of course use a separate MIDI
player/recorder, and route the events through mididings, but you'd lose all
meta information (tempo, time signature, etc.) from the original file.
Having said that, I don't really know a better
solution.
midish (http://caoua.org/midish/) can import/export MIDI files, but I'm not
sure how/if it can do the transformation you want.
I haven't check out yet.
PS. I join an
example with my post. If you open it in a sequencer and
look at the bassdrum you'll figure out what i say.
As far as I can see, all tracks in your MIDI file are on the same channel,
so CC #7 messages on one track will affect the other tracks as well. You
should probably assign a different channel to each track.
Cheers,
Dominic
Cheers,
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
11:07 p.m.
On Thursday 20 November 2008 23:49:46 Philippe Hezaine wrote:
Recording to a new track in the same Rosegarden session is actually a very
good idea, I didn't think of that... You wouldn't lose anything that way.
As for the transformation, something like this should do the trick (not
thoroughly tested):
from mididings import *
class Volume2Velocity:
def __init__(self):
self.vol = {}
def __call__(self, ev):
if ev.type_ == CTRL and ev.param == 7:
self.vol[ev.channel] = ev.value
return None
elif ev.type_ == NOTEON and ev.channel in self.vol:
ev.velocity = self.vol[ev.channel]
return ev
run(Call(Volume2Velocity()))
Yes, but then you need to do the transformation one track at a time. If you
play multiple tracks through the same MIDI port on the same channel, some
notes will get the wrong velocity (unless every note event is immediately
preceded by a volume change).
Cheers,
Dominic
I'm not
sure mididings is the right tool for the job. The Python code
to transform the events would be quite simple, but mididings doesn't
support reading/writing MIDI files (yet?). You could of course use a
separate MIDI player/recorder, and route the events through mididings,
but you'd lose all meta information (tempo, time signature, etc.) from
the original file.
Yes. I was sucessfull to loop a midi file with Jack and Rosegarden. The
outputs was routed through mididings and went back to Rg for a new
record on a new track. But it was not for this purpose and i didn't look
for the meta information.
But for now, if i have a copy about these informations may be i can try
again this experience? It isn't a piece of music, only a succession of
drums patterns. I can organize them to avoid different time signature... As far as I
can see, all tracks in your MIDI file are on the same
channel, so CC #7 messages on one track will affect the other tracks as
well. You should probably assign a different channel to each track.
No. Like i have said these are drums patterns written by Lilypond. After
the famous transformation i could join all the tracks in one if i want.
21 Nov
21 Nov
8:22 a.m.
Dominic Sacré a écrit :
On Thursday 20 November 2008 23:49:46 Philippe Hezaine
wrote:
Recording to a new track in the same Rosegarden session is actually a very
good idea, I didn't think of that... You wouldn't lose anything that way.
As for the transformation, something like this should do the trick (not
thoroughly tested):
from mididings import *
class Volume2Velocity:
def __init__(self):
self.vol = {}
def __call__(self, ev):
if ev.type_ == CTRL and ev.param == 7:
self.vol[ev.channel] = ev.value
return None
elif ev.type_ == NOTEON and ev.channel in self.vol:
ev.velocity = self.vol[ev.channel]
return ev
run(Call(Volume2Velocity()))
Yes, but then you need to do the transformation one track at a time. If you
play multiple tracks through the same MIDI port on the same channel, some
notes will get the wrong velocity (unless every note event is immediately
preceded by a volume change).
Cheers,
Dominic
Thanks too much for your time and your information.
Bingo! I think it's the first time we get from a Lilypond file a midi
file with velocities values. But for all you're right. Note that I
cannot get a clean file with a loop in the same Rosegarden's session. I
have freezes and even crashes. Hence i output the midi from Rg through
mididings -> recording in Muse.
The raw outcome is in attach.
1) You could see a little latency in the attacks.
2) I lose the Lilypond header
3) Strangely i can't open the new file with the drum editor in
Rosegarden though it is on channel 10. (meta information?)
Anyway, thanks so much.
Have fun.
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
I'm
not sure mididings is the right tool for the job. The Python code
to transform the events would be quite simple, but mididings doesn't
support reading/writing MIDI files (yet?). You could of course use a
separate MIDI player/recorder, and route the events through mididings,
but you'd lose all meta information (tempo, time signature, etc.) from
the original file.
Yes. I was sucessfull to loop a midi file with Jack and
Rosegarden. The
outputs was routed through mididings and went back to Rg for a new
record on a new track. But it was not for this purpose and i didn't look
for the meta information.
But for now, if i have a copy about these informations may be i can try
again this experience? It isn't a piece of music, only a succession of
drums patterns. I can organize them to avoid different time signature... As far as
I can see, all tracks in your MIDI file are on the same
channel, so CC #7 messages on one track will affect the other tracks as
well. You should probably assign a different channel to each track.
No. Like i
have said these are drums patterns written by Lilypond. After
the famous transformation i could join all the tracks in one if i want.
7:08 p.m.
On Friday 21 November 2008 13:27:18 Philippe Hezaine wrote:
Bingo! I think it's the first time we get from a
Lilypond file a midi
file with velocities values. But for all you're right. Note that I
cannot get a clean file with a loop in the same Rosegarden's session. I
have freezes and even crashes. Hence i output the midi from Rg through
mididings -> recording in Muse.
You probably need to disable "MIDI Thru Routing" in Rosegarden, otherwise
connecting Rg -> mididings -> Rg creates a MIDI feedback loop, and
Rosegarden doesn't like that...
The raw outcome is in attach.
This file shows the problem I mentioned, with multiple tracks on the same
channel. In your original MIDI file, the volume of the Hi-Hat and Congas is
constant, but in this one, the velocity varies. Are you sure that's what
you want?
1) You could see a little latency in the attacks.
You're using JACK transport to synchronize Rg and Muse?
I can reproduce this latency, and it appears to depend on my JACK buffer
size. The recording seems to be off by one period, but I don't know yet why
this happens, or whose fault it is.
If you can get recording in Rg to work this won't be an issue.
But note that even with JACK transport out of the equation, there will be a
very slight latency, because unfortunately mididings doesn't preserve the
exact event timestamps. You may want to quantize the recorded data.
Cheers,
Dominic
22 Nov
22 Nov
8:28 a.m.
Dominic Sacré a écrit :
No, of course. It's a testing file, with some "deliberately holes" just
to study the results. The CC7 are created by the dynamics in Lilypond.
There are 11 levels: from pppp (veloc = 43) to sf (veloc = 127). And
you can even reconfigure the equivalences if you want.
If you are needed i'll put on my site the Merengue's sources. If you
don't know Lilypond you can always look at the pdf.
Yes. And i've tried the recording in reverse order: Muse -> Rg. The
results are completely different. But i'm not sure it's a valuable
information. Perhaps i must check out my rtc configuration before.
On Friday 21 November 2008 13:27:18 Philippe Hezaine
wrote:
Unfortunatly it doesn't matter. I've tried different options and the
results are the same. For information, after a loop session in Rsgd,
when i quit (Ctrl-C) mididings i get this message:
/home/phil/python/mididings-CC7-to-velocity.py
^Cpython: /usr/include/boost/shared_ptr.hpp:315: T*
boost::shared_ptr<T>::operator->() const [with T = Backend]: Assertion
`px != 0' failed.
Abandon
I'm on Gentoo with pro-audio. Rosegarden 1.7.0. But all is done with my
"testing" computer and the onboard soundcard.
Otherwise with two sequencers, the console is clean.
Bingo! I think it's the first time we get
from a Lilypond file a midi
file with velocities values. But for all you're right. Note that I
cannot get a clean file with a loop in the same Rosegarden's session. I
have freezes and even crashes. Hence i output the midi from Rg through
mididings -> recording in Muse.
You probably need to disable "MIDI Thru Routing" in Rosegarden, otherwise
connecting Rg -> mididings -> Rg creates a MIDI feedback loop, and
Rosegarden doesn't like that...
The raw
outcome is in attach.
This file shows the problem I mentioned, with multiple tracks on the same
channel. In your original MIDI file, the volume of the Hi-Hat and Congas is
constant, but in this one, the velocity varies. Are you sure that's what
you want?
1) You could
see a little latency in the attacks.
You're using JACK transport to synchronize Rg and Muse? I can reproduce this latency, and it appears to depend
on my JACK buffer
size. The recording seems to be off by one period, but I don't know yet why
this happens, or whose fault it is.
In the next days i'll install mididings on my other computer with a
Multiface and i'll try again.
If you can get recording in Rg to work this won't
be an issue.
But note that even with JACK transport out of the equation, there will be a
very slight latency, because unfortunately mididings doesn't preserve the
exact event timestamps. You may want to quantize the recorded data.
Yes. It's not an issue for The Drummer's Free Art.
Cheers,
Dominic
All the best.
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
7:32 p.m.
On Saturday 22 November 2008 13:33:25 Philippe Hezaine wrote:
Yes. And i've tried the recording in reverse order: Muse -> Rg. The
results are completely different. But i'm not sure it's a valuable
information. Perhaps i must check out my rtc configuration before.
There was a bug in JACK transport that could cause an offset of one period.
It's supposedly fixed in recent versions (svn), but maybe there's still
something wrong in Muse or Rosegarden.
Cheers,
Dominic
Dominic Sacré a écrit :
I don't know then. Maybe talk to the Rosegarden guys? I'm sure this must be
possible somehow (it works for me, after all).
You probably need to disable "MIDI Thru
Routing" in Rosegarden,
otherwise connecting Rg -> mididings -> Rg creates a MIDI feedback
loop, and Rosegarden doesn't like that...
Unfortunatly it doesn't matter. I've tried different options and the
results are the same. For information, after a loop session in Rsgd,
when i quit (Ctrl-C) mididings i get this message:
/home/phil/python/mididings-CC7-to-velocity.py
^Cpython: /usr/include/boost/shared_ptr.hpp:315: T*
boost::shared_ptr<T>::operator->() const [with T = Backend]: Assertion
`px != 0' failed.
Abandon
Thanks, I uploaded a new version of mididings today that fixes this bug. If
it doesn't, please let me know. This is not relevant to the feedback loop
issue though.
1) You could see a little latency in the attacks.
You're using JACK transport to synchronize Rg and Muse?
23 Nov
23 Nov
1:28 p.m.
Dominic Sacré a écrit :
Thanks, I uploaded a new version of mididings today that fixes this bug. If
it doesn't, please let me know. This is not relevant to the feedback loop
issue though.
How do you uninstall the previous version? Is there some special way?
And perhaps my first install is not clean. I've all built in /usr/local
and only kept a copy of the examples in my /home which is the one with
the execution rights. Right?
In a other part, I've uploaded Rsgd to 1.7.2-r1. Same results.
The better i've got is the posted midi from the Rsgd -> Muse
configuration. IMHO, not too bad for an experience. I plan to try
Qtractor but i haven't it in hands yet.
Oh! please, note that it's not a sequencer's challenge.
Cheers,
Dominic
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
24 Nov
24 Nov
11:48 a.m.
On Sunday 23 November 2008 18:33:17 Philippe Hezaine wrote:
How do you uninstall the previous version? Is there
some special way?
Usually when you install a new version to the same prefix, the old files
will simply be overwritten.
To make sure the old version is really gone, you can go to the package
directory (something like /usr/lib/python2.5/site-packages) and remove the
mididings directory and the file _mididings.so.
And perhaps my first install is not clean. I've
all built in /usr/local
and only kept a copy of the examples in my /home which is the one with
the execution rights. Right?
Sounds alright.
By the way, try mididings-20081123. Reading and writing MIDI files turned
out to be much easier than I expected (thanks to libsmf), so I quickly
added support for processing MIDI files directly. It's not pretty and needs
more testing, but I tried it on your Merengue and it seems to do the right
thing (keeping all metadata intact).
Support for libsmf must be enabled explicitly at compile time
(./setup.py build --enable-smf).
You can use it by replacing the run() function with process_file():
process_file('infile.mid', 'outfile.mid', Call(Volume2Velocity()))
Cheers,
Dominic
12:20 p.m.
Dominic Sacré a écrit :
On Sunday 23 November 2008 18:33:17 Philippe Hezaine
wrote:
Fabulous! I'm going to try it quickly.
For the news i tested Rsgd -> mididings -> qtractor. Exactly the same
good results up to now. But surprisingly qtractor -> mididings -> Rsgd
give me all the velocities at 64. ( but it's with the old version. )
Cheers
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
How do you uninstall the previous version? Is
there some special way?
Usually when you install a new version to the same prefix, the old files
will simply be overwritten.
To make sure the old version is really gone, you can go to the package
directory (something like /usr/lib/python2.5/site-packages) and remove the
mididings directory and the file _mididings.so.
And perhaps my first install is not clean.
I've all built in /usr/local
and only kept a copy of the examples in my /home which is the one with
the execution rights. Right?
Sounds alright.
By the way, try mididings-20081123. Reading and writing MIDI files turned
out to be much easier than I expected (thanks to libsmf), so I quickly
added support for processing MIDI files directly. It's not pretty and needs
more testing, but I tried it on your Merengue and it seems to do the right
thing (keeping all metadata intact).
Support for libsmf must be enabled explicitly at compile time
(./setup.py build --enable-smf).
You can use it by replacing the run() function with process_file():
process_file('infile.mid', 'outfile.mid', Call(Volume2Velocity()))
Cheers,
Dominic
1:16 p.m.
Dominic Sacré a écrit :
By the way, try mididings-20081123. Reading and writing MIDI files turned
out to be much easier than I expected (thanks to libsmf), so I quickly
added support for processing MIDI files directly. It's not pretty and needs
more testing, but I tried it on your Merengue and it seems to do the right
thing (keeping all metadata intact).
Support for libsmf must be enabled explicitly at compile time
(./setup.py build --enable-smf).
You can use it by replacing the run() function with process_file():
process_file('infile.mid', 'outfile.mid', Call(Volume2Velocity()))
Cheers,
Dominic
I get this with ./setup.py build --enable-smf
....
i686-pc-linux-gnu-g++ -pthread -fPIC -DENABLE_JACK_MIDI=1 -DENABLE_SMF=1
-I/usr/include/alsa -Isrc -I/usr/include/python2.5 -c
-DENABLE_JACK_MIDI=1 -DENABLE_SMF=1 -I/usr/include/alsa -Isrc
-I/usr/include/python2.5 -c src/engine.cc -o
build/temp.linux-i686-2.5/src/engine.o
Dans le fichier inclus à partir de src/engine.cc:20:
src/backend_smf.hh:21:17: erreur: smf.h : Aucun fichier ou répertoire de
ce type
src/backend_smf.hh:39: erreur: ‘smf_t’ was not declared in this scope
src/backend_smf.hh:39: erreur: patron de l'argument 1 est invalide
src/backend_smf.hh:40: erreur: ‘smf_t’ was not declared in this scope
src/backend_smf.hh:40: erreur: patron de l'argument 1 est invalide
src/backend_smf.hh: In member function ‘virtual size_t
BackendSmf::num_out_ports() const’:
src/backend_smf.hh:36: erreur: base operand of ‘->’ is not a pointer
error: command 'i686-pc-linux-gnu-g++' failed with exit status 1
I've qjackctl --version
Qt: 4.3.3
QjackCtl: 0.3.3
and
jack-smf-utils-1.0 USE="readline -lash
May be i've just missed a library?
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
2:49 p.m.
On Monday 24 November 2008 18:21:14 Philippe Hezaine wrote:
I get this with ./setup.py build --enable-smf
....
i686-pc-linux-gnu-g++ -pthread -fPIC -DENABLE_JACK_MIDI=1 -DENABLE_SMF=1
-I/usr/include/alsa -Isrc -I/usr/include/python2.5 -c
-DENABLE_JACK_MIDI=1 -DENABLE_SMF=1 -I/usr/include/alsa -Isrc
-I/usr/include/python2.5 -c src/engine.cc -o
build/temp.linux-i686-2.5/src/engine.o
Dans le fichier inclus à partir de src/engine.cc:20:
src/backend_smf.hh:21:17: erreur: smf.h : Aucun fichier ou répertoire de
ce type
src/backend_smf.hh:39: erreur: ‘smf_t’ was not declared in this scope
src/backend_smf.hh:39: erreur: patron de l'argument 1 est invalide
src/backend_smf.hh:40: erreur: ‘smf_t’ was not declared in this scope
src/backend_smf.hh:40: erreur: patron de l'argument 1 est invalide
src/backend_smf.hh: In member function ‘virtual size_t
BackendSmf::num_out_ports() const’:
src/backend_smf.hh:36: erreur: base operand of ‘->’ is not a pointer
error: command 'i686-pc-linux-gnu-g++' failed with exit status 1
I've qjackctl --version
Qt: 4.3.3
QjackCtl: 0.3.3
and
jack-smf-utils-1.0 USE="readline -lash
May be i've just missed a library?
jack-smf-utils includes libsmf, but doesn't seem to install it for other
programs to use. You'll need the standalone version of libsmf-1.1. You can
run "pkg-config --modversion smf" to check if it's installed correctly.
Dominic
4:42 p.m.
Dominic Sacré a écrit :
jack-smf-utils includes libsmf, but doesn't seem to install it for other
programs to use. You'll need the standalone version of libsmf-1.1. You can
run "pkg-config --modversion smf" to check if it's installed correctly.
pkg-config --modversion smf
Package smf was not found in the pkg-config search path.
Perhaps you should add the directory containing `smf.pc'
to the PKG_CONFIG_PATH environment variable
No package 'smf' found
Do i have to install it manually? I can dowload the sources and put them
in /usr/portage/distfiles and then emerge but what do i do with
PKG_CONFIG_PATH ?
Regards
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
5:18 p.m.
On Monday 24 November 2008 21:47:41 Philippe Hezaine wrote:
pkg-config --modversion smf
Package smf was not found in the pkg-config search path.
Perhaps you should add the directory containing `smf.pc'
to the PKG_CONFIG_PATH environment variable
No package 'smf' found
Do i have to install it manually? I can dowload the sources and put them
in /usr/portage/distfiles and then emerge but what do i do with
PKG_CONFIG_PATH ?
If there's no Gentoo package for libsmf, it's probably easiest to install it
the usual, distro-independent way, i.e. extract the tarball and run
"./configure && make && make install".
That will install libsmf to /usr/local. If it's still not found after that,
try "export PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig"
before building mididings.
Dominic
5:46 p.m.
Dominic Sacré a écrit :
If there's no Gentoo package for libsmf, it's
probably easiest to install it
the usual, distro-independent way, i.e. extract the tarball and run
"./configure && make && make install".
That will install libsmf to /usr/local. If it's still not found after that,
try "export PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig"
before building mididings.
Dominic
And Bingo again, Dominic.
Many thanks.
See you later.
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
7:30 p.m.
Philippe Hezaine a écrit :
And Bingo again, Dominic.
Many thanks.
See you later.
Champagne!
We get for the first time in the world a midi file with velocities from
a lilypond file!
Here is the result after i've merged the 11 original tracks in one.
The transformation is perfect! Of course it's an experimental file.
Many thanks to Dominic who is going to step up the Drummer's Free Art in
a new land.
Dominic, please make it a point of honour to announce this great news to
the lilypond's devs. Do you want?
Have fun.
--
Phil.
Superbonus-Project (Site principal) <http://superbonus.project.free.fr>
Superbonus-Project (Plate-forme d'échange):
<http://philippe.hezaine.free.fr>
25 Nov
25 Nov
7:52 p.m.
On Tuesday 25 November 2008 00:35:58 Philippe Hezaine wrote:
Champagne!
We get for the first time in the world a midi file with velocities from
a lilypond file!
Great! Glad I could help.
Dominic, please make it a point of honour to announce
this great news to
the lilypond's devs. Do you want?
I'll leave that to you, after all you're the one actually using it to make
music :)
Cheers,
Dominic
5875
days inactive
5880
days old
linux-audio-user@lists.linuxaudio.org
17 comments
2 participants
participants (2)
-
Dominic Sacré -
Philippe Hezaine