13 Oct
2002
13 Oct
'02
3:31 p.m.
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
FreqTweak is a tool for FFT-based realtime audio spectral manipulation
and display. It provides several algorithms for processing audio data
in the frequency domain and a highly interactive GUI to manipulate the
associated filters for each. It also provides high-resolution spectral
displays in the form of scrolling- raster spectragrams and energy vs
frequency plots displaying both pre- and post-processed spectra.
It currently relies on JACK for low latency audio interconnection and
delivery. Thus, it is only supported on Linux.
FreqTweak is an extremely addictive audio toy, I have to pry myself
away from playing with it so I can work on it! I hope it has value
for serious audio work too (sound design, etc). The spectrum analysis
is pretty useful in its own right.
FreqTweak supports manipulating the spectral filters at several
frequency resolutions (64,128,256,512,1024, or 2048 bands) depending
on your needs and resources. Overlap and windowing are also
selectable.
The GUI filter graph manipulators (and analysis plots) have selectable
frequency scale types: 1x and 2x linear, and two log scales to help
with modulating the musical frequencies. Filters can be linked across
multiple channels. The plots are resizable and zoomable (y-axis) to
allow precise editing of filter values.
The current processing filters are described below in the order audio
is processed in the chain. Any or all of the filters can be
bypassed. The state of all filters can be stored or loaded as presets.
Spectral Analysis -- Multicolor scrolling-raster spectragram,
or energy vs. freq line or bar plots... one shows
pre-processed, another shows post-processed.
EQ -- Your basic multi-band frequency attenuation. But you get
an unhealthy number of bands...
Pitch Scaling -- This is an interesting application of
Sprengler's pitch scaling algorithm (used in Steve Harris'
LADSPA plugin). If you keep all the bins at the same scale, it
is equivalent to Steve's plugin, but when you start applying
different scales per frequency bin, things quickly get weird.
Gate -- This is a double filter where a given frequency band is
allowed to pass through (unaltered) if the power on that band
is between two dB thresholds... otherwise its gain is clamped
to 0.
Delay -- This lets you delay the audio on a per frequency-bin
basis yielding some pretty wild effects (or subtle, if you are
careful). A feedback filter controls the feedback of the delay
per bin (be careful with this one). This is basically what
Native Instrument's Spektral-Delay accomplishes. Granted, I
don't have all the automated filter modulations (yet ;). See
their website for audio examples of what is possible with this
cool effect.
Have fun... report bugs...
Jesse Chappell <jesse(a)essej.net>
13 Oct
13 Oct
4:36 p.m.
Jesse--
This project looks extremely cool! I will have to check it out when I
get home to my linux computer (I'm presently remote with my notebook --
a powerbook with no Linux partition).
My question is this: how modular is the audio code (i.e. is it
possible to support different audio targets at the makefile level)? It
is, as you likely know, fairly easy to port between various
callback-based libraries, and I'd be interested in taking a crack at
making a CoreAudio port for OS X. (Since there is a wxWindows port for
OS X, the "hard part" is not so hard. :-)
best,
wb
6:53 p.m.
My question is this: how modular is the audio code (i.e. is it
possible to support different audio targets at the makefile level)? It is,
as you likely know, fairly easy to port between various
callback-based libraries, and I'd be interested in taking a crack at making a
CoreAudio port for OS X. (Since there is a wxWindows port for OS X, the
"hard part" is not so hard. :-)
It should be pretty easy... I planned to support alternate I/O subsytems, it
should just be a matter of subclassing FTioSupport ( as FTjackSupport is) and
implementing it. Hah! Wouldn't we be so lucky :)
jlc
8:18 p.m.
On Sun, 13 Oct 2002 16:49:12 -0400 (EDT)
"Jesse Chappell" <jesse(a)essej.net> wrote:
An even better solution for the long-term would be to use PortAudio
(www.portaudio.com), which will give you Windows, ALSA, JACK, OSS, CoreAudio,
ASIO, and more. Basically it would be like your current strategy, except all
the subclasses have already been written and tested for you! :-)
The current stable release doesn't support JACK or ALSA, but the development
branch does (I can speak for these since I wrote them :-). So I wouldn't
suggest moving to PortAudio right now since JACK support is a priority for you,
but you may want to consider it for the long-term.
Josh
--
"Only Angels can sing that high, and only dogs can hear it." -- James Evans
My question is
this: how modular is the audio code (i.e. is it
possible to support different audio targets at the makefile level)? It is,
as you likely know, fairly easy to port between various
callback-based libraries, and I'd be interested in taking a crack at making a
CoreAudio port for OS X. (Since there is a wxWindows port for OS X, the
"hard part" is not so hard. :-)
It should be pretty easy... I planned to support alternate I/O subsytems, it
should just be a matter of subclassing FTioSupport ( as FTjackSupport is) and
implementing it. Hah! Wouldn't we be so lucky :) 
5 p.m.
Jesse,
This looks like a lot of fun. If I wasn't having an open house in a hour,
I'd start trying to build it right now. (Let's see how it stands up to a
newbie like me trying to build it, right?!) ;-)
I had been looking at the NI tool and thinking about it. Now I get to
play in the right OS! Thanks!
FYI - I could not get the wxWindows link to work for me.
Very cool looking tool set. I look forward to this evening.
Cheers,
Mark
-----Original Message-----
From: linux-audio-dev-admin(a)music.columbia.edu
[mailto:linux-audio-dev-admin@music.columbia.edu]On Behalf Of Jesse
Chappell
Sent: Sunday, October 13, 2002 10:27 AM
To: linux-audio-user(a)music.columbia.edu
Cc: linux-audio-dev(a)music.columbia.edu;
linux-audio-announce(a)music.columbia.edu;
jackit-devel(a)lists.sourceforge.net
Subject: [linux-audio-dev] Announcing initial release of FreqTweak
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
FreqTweak is a tool for FFT-based realtime audio spectral manipulation
and display. It provides several algorithms for processing audio data
in the frequency domain and a highly interactive GUI to manipulate the
associated filters for each. It also provides high-resolution spectral
displays in the form of scrolling- raster spectragrams and energy vs
frequency plots displaying both pre- and post-processed spectra.
It currently relies on JACK for low latency audio interconnection and
delivery. Thus, it is only supported on Linux.
FreqTweak is an extremely addictive audio toy, I have to pry myself
away from playing with it so I can work on it! I hope it has value
for serious audio work too (sound design, etc). The spectrum analysis
is pretty useful in its own right.
FreqTweak supports manipulating the spectral filters at several
frequency resolutions (64,128,256,512,1024, or 2048 bands) depending
on your needs and resources. Overlap and windowing are also
selectable.
The GUI filter graph manipulators (and analysis plots) have selectable
frequency scale types: 1x and 2x linear, and two log scales to help
with modulating the musical frequencies. Filters can be linked across
multiple channels. The plots are resizable and zoomable (y-axis) to
allow precise editing of filter values.
The current processing filters are described below in the order audio
is processed in the chain. Any or all of the filters can be
bypassed. The state of all filters can be stored or loaded as presets.
Spectral Analysis -- Multicolor scrolling-raster spectragram,
or energy vs. freq line or bar plots... one shows
pre-processed, another shows post-processed.
EQ -- Your basic multi-band frequency attenuation. But you get
an unhealthy number of bands...
Pitch Scaling -- This is an interesting application of
Sprengler's pitch scaling algorithm (used in Steve Harris'
LADSPA plugin). If you keep all the bins at the same scale, it
is equivalent to Steve's plugin, but when you start applying
different scales per frequency bin, things quickly get weird.
Gate -- This is a double filter where a given frequency band is
allowed to pass through (unaltered) if the power on that band
is between two dB thresholds... otherwise its gain is clamped
to 0.
Delay -- This lets you delay the audio on a per frequency-bin
basis yielding some pretty wild effects (or subtle, if you are
careful). A feedback filter controls the feedback of the delay
per bin (be careful with this one). This is basically what
Native Instrument's Spektral-Delay accomplishes. Granted, I
don't have all the automated filter modulations (yet ;). See
their website for audio examples of what is possible with this
cool effect.
Have fun... report bugs...
Jesse Chappell <jesse(a)essej.net>
7:38 p.m.
New subject: [linux-audio-dev] Re: [Jackit-devel] Announcing initial release of FreqTweak
Hi,
Jesse Chappell hat gesagt: // Jesse Chappell wrote:
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
FreqTweak is an extremely addictive audio toy, I have to pry myself
away from playing with it so I can work on it!
Darn, you're right. I'm running my Pd-patches through FreqTweak for
hours now ;)
This is the the most impressive, ahem, effect box (?) I've seen on
Linux yet. It's similar to NI's Spektral Delay, but that makes it
even more useful. This is, what Jack must have been designed for.
Thank you a lot.
("Must turn volume higher, cannot stop yet ...")
ciao
--
Frank Barknecht _ ______footils.org__
8:54 p.m.
New subject: [linux-audio-dev] Re: [Jackit-devel] Announcing initial release of FreqTweak
On Sun, Oct 13, 2002 at 01:26:48 -0400, Jesse Chappell wrote:
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
This is /really/ cool.
Could you add a 4096 bin mode? I've got some great burbling noises coming
out at 2048, but I think the extra res. would make it even better.
I did once manage to get FT disconnected from jack when changing bin
sizes, but only once out of a few dozen changes, so it might be unrealted.
Cheers,
Steve
10:55 p.m.
New subject: [linux-audio-dev] Re: [Jackit-devel] Announcing initial release of FreqTweak
Steve,
Could you add a 4096 bin mode? I've got some great
burbling noises coming out
at 2048, but I think the extra res. would make it even better.
Check out the latest CVS... I threw it in for you ;)
Memory usage is a bit higher (but not by much...).
I did once manage to get FT disconnected from jack
when changing bin sizes,
but only once out of a few dozen changes, so it might be unrealted.
Yes, I believe there are some buffer overrun gremlins hiding in there that
really only come out at those big FFT sizes. I find that I lose too much
time-domain resolution at those sizes for anything with transients. It just
mushes right up.
jlc
14 Oct
14 Oct
6:31 a.m.
New subject: [linux-audio-dev] Re: [Jackit-devel] Announcing initial release of FreqTweak
On Sun, Oct 13, 2002 at 08:50:25 -0400, Jesse Chappell wrote:
That was the effect I was after. The mushy freq domain effect from small
FFT sizes can be good too.
- Steve
Check out the latest CVS... I threw it in for you ;)
Cheers :)
One more freq domain effect you might like to add is a bin remapper, it
allows you to put the amplitude and phase information from any input bin
into any output bin, the UI might be tricky, but you could do it with a
sqaure graph, like a matrix mixer.
I did once
manage to get FT disconnected from jack when changing bin sizes,
but only once out of a few dozen changes, so it might be unrealted.
Yes, I believe there are some buffer overrun gremlins hiding in there that
really only come out at those big FFT sizes. I find that I lose too much
time-domain resolution at those sizes for anything with transients. It just
mushes right up. 
13 Oct
13 Oct
9:04 p.m.
On Sun, 2002-10-13 at 19:26, Jesse Chappell wrote:
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
This is a really impressive piece of software. You did an excellent job,
and the sound is just amazing. FFT is real good when processing stereo
signals as it keeps phase relationships between both channels, and I
love the twisted but somehow natural sounds it produces.
I've been playing with FreqTweak for a couple of hours now, and it
segfaulted several time, most notably with large window sizes.
The preset "load" button causes a segfault if the listbox is empty.
Here's a quick wishlist:
- being able to use the under-the-threshold part of the signal in the
gate module, a la Logic Audio Spectral Gate
- being able to apply pitch modifications within each pass of the delay
feedback (maybe this one is total sci-fi as it would imply a recursive
behaviour, but I think the resulting sounds could be really interesting)
- being able to undock each module in its own window.
- being able to extend the maximum delay time (more than 10 secs would
be nice).
Thanks for this great app, Jesse !
Regards,
Thomas
11:11 p.m.
Thomas wrote:
I've been playing with FreqTweak for a couple of
hours now, and it
segfaulted several time, most notably with large window sizes.
The preset "load" button causes a segfault if the listbox is empty.
Thanks for the load button tip, it is now fixed in the latest CVS.
The segfaults at large FFT sizes are buffer overrun gremlins, I'm sure...
gdb hasn't helped yet on those.
Here's a quick wishlist:
- being able to use the under-the-threshold part of
the signal in the gate
module, a la Logic Audio Spectral Gate
I need to check it out...
- being able to apply pitch modifications within each
pass of the delay
feedback (maybe this one is total sci-fi as it would imply a recursive
behaviour, but I think the resulting sounds could be really interesting)
I'll have to think about this! It hurts my brain...
- being able to undock each module in its own window.
Yeah, the modules could use some more "modularity" all the way around.
Reordering, total minimization, popoff, etc. I seem to be running out of
mouse-button-modifiers too, it is already pretty absurd.
- being able to extend the maximum delay time (more
than 10 secs would be
nice).
This is no problem.... I don't want it to be _too_ much of a memory hog, but
selectable delay times are nice. I would probably deactivate - realloc -
activate as a tradeoff (that way I can support as much delay as you have memory)
at the expense of interrupted audio on transitions.
Thanks for this great app, Jesse !
My pleasure :)
jlc
14 Oct
14 Oct
6:49 a.m.
On Sun, Oct 13, 2002 at 09:06:34PM -0400, Jesse Chappell wrote:
Just feed the output of the feedback into a part of the chain before the
pitch scale. Infact being able to select the destination of the
feedback, combined with reordering would rule.
- Steve
- being able
to apply pitch modifications within each pass of the delay
feedback (maybe this one is total sci-fi as it would imply a recursive
behaviour, but I think the resulting sounds could be really interesting)
I'll have to think about this! It hurts my brain... 
11:21 a.m.
On Sun, 2002-10-13 at 19:26, Jesse Chappell wrote:
> Announcing the initial release of FreqTweak (v0.4)
>
> http://freqtweak.sourceforge.net
[ ... ]
Here's a quick wishlist:
drat. the ardour project loses one of its more capable programmers :)
--p
6:53 a.m.
Jesse Chappell wrote:
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
hi jesse ! after all the enthusiastic feedback i want to try it out,
too.
unfortunately, the sf downloads of either 0.4.0, .1 and .2 bail out
after about 2k, and the reconf script in cvs complains about a missing
acconfig - i have never seen that command before. it generates a
./configure script, which then makes makefiles, but the compilation
stops with "missing separator in Makefile"...
any hints ?
regards,
jörn
--
Jörn Nettingsmeier
Kurfürstenstr 49, 45138 Essen, Germany
http://spunk.dnsalias.org (my server)
http://www.linuxdj.com/audio/lad/ (Linux Audio Developers)
2:17 p.m.
hi jesse ! after all the enthusiastic feedback i want to try it out, too.
unfortunately, the sf downloads of either 0.4.0, .1 and .2 bail out after
about 2k, and the reconf script in cvs complains about a missing acconfig - i
have never seen that command before. it generates a
./configure script, which then makes makefiles, but the compilation stops with
"missing separator in Makefile"...
Don't know what the deal with SF is... it works for me.
About the 'reconf' script.... I believe I am using a terribly outdated method of
using autoconf and friends. You will need to download autotools-0.10 (found
at http://www.amath.washington.edu/~lf/tutorials/autoconf/ ) which contains
acconfig.
I just downloaded the GNU Autoconf, Automake, and Libtool book and I hope to
rework it soon.
jlc
15 Oct
15 Oct
5:28 a.m.
Jesse Chappell wrote:
argh. it's the *link* to a download page... ok, silly me. i just saved a
html page and wondered why tar barfs.
but tell me, how on earth can you have a link like "filename.tar.gz"
which is not a tarball.... mysterious are the ways of sf.
thanks anyway, i'm building it later today...
best,
jörn
--
Jörn Nettingsmeier
Kurfürstenstr 49, 45138 Essen, Germany
http://spunk.dnsalias.org (my server)
http://www.linuxdj.com/audio/lad/ (Linux Audio Developers)
hi jesse ! after all the enthusiastic feedback i want to try it out, too.
unfortunately, the sf downloads of either 0.4.0, .1 and .2 bail out after
about 2k, and the reconf script in cvs complains about a missing acconfig - i
have never seen that command before. it generates a
./configure script, which then makes makefiles, but the compilation stops with
"missing separator in Makefile"...
Don't know what the deal with SF is... it works for me.
7:05 a.m.
Hi Joern,
Jesse Chappell wrote:
argh. it's the *link* to a download page... ok, silly me. i just saved a
html page and wondered why tar barfs.
but tell me, how on earth can you have a link like "filename.tar.gz"
which is not a tarball.... mysterious are the ways of sf.
thanks anyway, i'm building it later today...
I keep stepping over that often
too. Just click onto it. The set the proper
content-type, so the browser will ask you to save it. B.T.W. you can call your
links what ever you like, it's the content-type of the http-protocol that
matters. It is just that most web-servers use the extension to set the
content-type automaticaly (when delivering the content from a file).
Stefan
--
\|/ Stefan Kost
<@ @> private business
+-oOO-(_)-OOo------------------------------------------------------ - - - - -
| __ Address Simildenstr. 5 HTWK Leipzig, Fb IMN, Postfach 300066
| /// 04277 Leipzig 04277 Leipzig
| __ /// Germany Germany
| \\\/// Phone +49341 2253538 +49341 30766101
| \__/ EMail st_kost(a)gmx.net kost(a)imn.htwk-leipzig.de
| WWW www.sonicpulse.de www.imn.htwk-leipzig.de/~kost/about.html
===-=-=--=---=---------------------------------- - - - - -
hi jesse !
after all the enthusiastic feedback i want to try it out, too.
unfortunately, the sf downloads of either 0.4.0, .1 and .2 bail out after
about 2k, and the reconf script in cvs complains about a missing acconfig - i
have never seen that command before. it generates a
./configure script, which then makes makefiles, but the compilation stops with
"missing separator in Makefile"...
Don't know what the deal with SF is... it works for me.
2:04 p.m.
Stefan Kost wrote:
Hi Joern,
> Jesse Chappell wrote:
>
>>>hi jesse ! after all the enthusiastic feedback i want to try it out, too.
>>>unfortunately, the sf downloads of either 0.4.0, .1 and .2 bail out after
>>>about 2k,
>>Don't know what the deal with SF is... it
works for me.
>>
> argh. it's the *link* to a download page...
ok, silly me. i just saved a
> html page and wondered why tar barfs.
> but tell me, how on earth can you have a link like "filename.tar.gz"
> which is not a tarball.... mysterious are the ways of sf.
> thanks anyway, i'm building it later today...
I keep stepping over that often too. Just click onto
it. The set the proper
content-type, so the browser will ask you to save it. B.T.W. you can call your
links what ever you like, it's the content-type of the http-protocol that
matters.
sure. it's just that when i see a filename that is a link, i click "save
as" by default...
maybe i need a firmware patch to evaluate mime types :)
--
Jörn Nettingsmeier
Kurfürstenstr 49, 45138 Essen, Germany
http://spunk.dnsalias.org (my server)
http://www.linuxdj.com/audio/lad/ (Linux Audio Developers)
3:02 p.m.
Jesse Chappell wrote:
Announcing the initial release of FreqTweak (v0.4)
http://freqtweak.sourceforge.net
oh my gawd...
this is the sickest^H^H^H^H^H^H^Hgreatest thing since, well, i don't
know. i need a week off to play with it....
in fact, i've been looking for a thing like this for two years for our
linuxtag booth to attract people. it has lots of visual appeal, and you
can do weird stuff in real-time. yeah.
frank, what do you think ? how will we piss off the php guys in the
booth next to us in 2003 ? >:->
hats off jesse !
--
Jörn Nettingsmeier
Kurfürstenstr 49, 45138 Essen, Germany
http://spunk.dnsalias.org (my server)
http://www.linuxdj.com/audio/lad/ (Linux Audio Developers)
3:38 p.m.
Hi,
Jörn wrote:
[..]
in fact, i've been looking for a thing like this
for two years for our
linuxtag booth to attract people. it has lots of visual appeal, and you
can do weird stuff in real-time. yeah.
frank, what do you think ? how will we piss off the php guys in the
booth next to us in 2003 ? >:->
I've been on vacation until yesterday, and have just found this tool
today; will probably try it out at home tonight.
I'm sure we will have lots of noises to scare anyone away who comes too
close :-).
Eager to get home,
Frank
PS: Isn't it about time for a new release of Jack? I'm still using 0.34,
which is the so far only "packaged" release, but I keep on reading about
0.37, 0.38 (Ok, I admit this is the wrong list to ask ;-).
7:17 p.m.
Hi,
Frank Neumann hat gesagt: // Frank Neumann wrote:
PS: Isn't it about time for a new release of Jack?
I'm still using 0.34,
which is the so far only "packaged" release, but I keep on reading about
0.37, 0.38 (Ok, I admit this is the wrong list to ask ;-).
You're using the wrong distribution :)
Debian already has 0.38 packaged, thanks to Stefan Schwandter.
ciao
--
Frank Barknecht _ ______footils.org__
16 Oct
16 Oct
6:45 a.m.
Hi,
Frank Barknecht wrote:
Actually, I _am_ using Debian :-), though I don't do apt-get stuff, but
instead get all my packages "manually" (don't ask).
Still, this is very inconsistent then. I just double-checked, and
_according to the Jack website_, the most current release is still 0.34.
How can someone name a debianized packaged "0.38", when there is no
upstream 0.38 source archive? This is not good. He can call it
"jackit_cvs2002mmdd-v.deb" or similar if he wants, but calling it 0.38
is very confusing here.
Frank
--
Frank Neumann (Frank.Neumann(a)st.com), VIONA Development Center
STMicroelectronics, Karlstraße 27, 76133 Karlsruhe
PS: Isn't
it about time for a new release of Jack? I'm still using 0.34,
which is the so far only "packaged" release, but I keep on reading about
0.37, 0.38 (Ok, I admit this is the wrong list to ask ;-).
You're using the wrong distribution :)
Debian already has 0.38 packaged, thanks to Stefan Schwandter. 
12:54 p.m.
On Wed, 16 Oct 2002, Frank Neumann wrote:
Sortof, but then again the version numbers correspond more-or-less
exactly with cvs (any change and the version is bumped). So really
Stefan was doing us a favor, considering no one else has done distro
efforts. I'd step up to the plate, so to speak, but I'm not going to be
around any more after Friday.
So yes, an 'official' release would be nice. But barring that, what
Stefan did is nice, too.
regards,
wingo.
You're
using the wrong distribution :)
Debian already has 0.38 packaged, thanks to Stefan Schwandter.
Actually, I _am_ using Debian :-), though I don't do apt-get stuff, but
instead get all my packages "manually" (don't ask).
Still, this is very inconsistent then. I just double-checked, and
_according to the Jack website_, the most current release is still 0.34.
How can someone name a debianized packaged "0.38", when there is no
upstream 0.38 source archive? This is not good. He can call it
"jackit_cvs2002mmdd-v.deb" or similar if he wants, but calling it 0.38
is very confusing here.
8064
days inactive
8067
days old
linux-audio-dev@lists.linuxaudio.org
23 comments
14 participants
participants (14)
-
Andy Wingo -
David Gerard Matthews -
Frank Barknecht -
Frank Barknecht -
Frank Neumann -
Jesse Chappell -
Joern Nettingsmeier -
Joshua Haberman -
Mark Knecht -
Paul Davis -
Stefan Kost -
Steve Harris -
Thomas Charbonnel -
Will Benton