I was in the need for ffmpeg to output directly to JACK, which is not
(It has JACK input support, as in recording something from JACK and feed
it to something else, but not the other way around)
After trying a couple of methods that didn't work so great, like fifos
or bridging through alsa loopback, I just went ahead and implemented
this part in FFmpeg.
You can find the relevant code at
Not yet submitted upstream, still doing some testing.
If this is something useful for you, feedback / comments on the code is
The relevant part sits at this file
Have a great weekend!
Summary of this message:
* systemd got udev rules with new database for firewire node/unit
* then fw character device for audio is owned by 'audio' group with ACL
* the entries are added by my investigation, thus doesn't cover all
* if you have firewire audio devices not listed in README of below
repository, please contact to me with image of configuration ROM:
The way to create image file of configuration ROM is typically:
$ cat /sys/bus/firewire/devices/fw1/config_rom > filename.img
Here, I presuppose that Linux FireWire subsystem detects your device as
Well, in the past, access permission of Linux firewire character device
is decided by udev rules just for video devices. This was
inconvenient some project such as ALSA and FFADO to produce audio
The source code of libffado includes own file for udev rules to take
firewire character device owned by 'audio' group. Additionally the rules
gives 'ID_FFADO' tag, and systemd includes another udev rule to ACL
at logging-in time according to it.
As a whole, the above is not comprehensive and self-contained. I
proposed patchset to systemd for better solution and today it was merged.
In the patchset, I add some udev rules, based on hwdb for new entries of
node and units in IEEE 1394 bus. You can see the database.
The entries of database have below variables when matching to either
node or unit devices:
The added udev rules interpret the content of variables and decide group
owner of fw character device(see ). Furthermore, the variables are
used again to decide ACL in logging-in time(see , too).
The entries of database also include below variables:
They are expected to use applications such as PipeWire and PulseAudio for
better names of sound device, which binds to unit instead of node. I
expect the variables can obsolete my former patch for pulseaudio.
I handy write the entries of database from my investigation, thus
it could includes the lack of your device, or mistakes. I wish you to
contact to me with image file of configuration ROM when you can not find
your device in README of my collection repository, or when you find
any mistakes in database file.
Thanks for your cooperation in advance.
 4 rules in 'rules.d/50-udev-default.rules'
 many rules in 'libffado/60-ffado.rules'
 udev: use ID_MODEL/ID_VENDOR to give friendly name for FireWire devices
I'm new to Travis (and CI in general), but I'm currently up against the
That is, the Travis VM is failing to start JACK because it can't find an
audio device. I need JACK running in order to test my API.
Has anyone come across this before, or have any recommendations?
I've decided to release Drops.
Drops is a single audio file sample player plugin in lv2 and vst
format for linux.
Load an audio file, play it, loop it, pitch shift it, and make it into
something entirely new.
The sample engine is sfizz.
Drops is free software: you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation, either version 3 of the License, or any later
Some of the features:
- audio disk streaming, allowing for large audio files to be played
- waveform display with zooming and scrolling
+ note: use mousewheel to zoom in and out
- 'minimap' with scroll area
- sample in/out and loop points
- ADSR for amp, filter and pitch
- LFO, free running or sync to transport and bpm for amp, filter and pitch
source code and binaries (for linux)
I hope this plugin will be of use to you.
Rob van den Berg
I remember listening to the talk of researchers who were traveling to
different old cathedrals, particularly to Hagia Sophia in Turkey, and
measuring echo in these cathedrals. Such buildings add a lot of deep and
very prolonged echo which depends on the building's shape and materials.
They were quantifying the noise response too.
Are there LV2 plugins that can add same or similar echo as cathedrals add?
Please refer to this thread on linuxmusicians.com forum, starting here:
re. Non-Session-Manager fork descalation
I'm not willing to open and engage into yet another discussion about the
so called "NSM fork", though I still don't understand what jackaudio.org
has to do with the new(er)-session-management (aka "NSM fork") and why
it is now moved from linuxaudio.org to jackaudio.org.
I don't like it and as I said before (on #jack irc), it won't get my
vote in: please don't make the same mistake again and keep jackaudio.org
away from this diatribe.
So please, host it on its own website, or github.com page, or on
kx.studio or laborejo.org, as these are the proper to the main two
proponents to "the fork".
Thanks for listening
rncbc aka. Rui Nuno Capela
>please read the subject line!
>it's about the new-session-management move to jackaudio.org and only.
>it is NOT about the merits, technical or otherwise, of session managers in general nor NSM in particular.
Then let me rephrase:
Why should NSM not be hosted under jackaudio, since it is the session manager for JACK programs?
The response to that was "because NSM is technically independent of JACK" and my answer was in the previous mail.