14 Mar
2019
14 Mar
'19
5:38 p.m.
Hello!
I'm attempting to produce a set of sound files that will be loaded into
a sampler. Given a MIDI file that contains a single channel containing
eight one-second notes, I'd like to be able to instantiate a VST plugin,
load a preset into it, pass the MIDI file through it, and record the
output to an audio file without any leading or trailing silence. I'd
like to be able to do that entirely from the command line without any
interactivity. I have a couple of hundred repetitions of this that I'd
want to perform, so as you can imagine I don't want to waste any time
on manual input.
Is there anything open-source out there that allows for this kind of
headless, command-line rendering?
--
Mark Raynsford | http://www.io7m.com
14 Mar
14 Mar
6:01 p.m.
Mar 14 2019, Mark Raynsford has written:
...
Given a MIDI file that contains a single channel
containing
eight one-second notes, I'd like to be able to instantiate a VST plugin,
load a preset into it, pass the MIDI file through it, and record the
output to an audio file without any leading or trailing silence.
...
Hi Mark,
Csound can do it, but you also need the VST4CS opcodes, that are
currently provided separately in binary form, since the issue with the
VST_sdk
Admittedly, Csound is not the easiest - nor the most devilishly complex
- tool to use. I can't supply code to do it right here and now, but:
1. Csound can read a score from a MIDI file
2. Csound can output to a lot of audio file formats (if necessary
specified on the commandline)
3. The vstinit
http://www.csounds.com/manual/html/vstinit.html
vst_info, vstnote, vstparamset, vstparamget and vstprogset documentation
still available on this version of the Csound manual:
http://www.csounds.com/manual/html/
should contain enough of an example to move in the right direction.
Another idea: if your setup allows to load one specific VST plugin with
some GUI program, if that is enough, you could choose your MIDI player
and audio recorder on the commandline. There are enough of those, that
could even work together, using some kind of transport, mostly MMC/MTC,
some JACK transport.
Best wishes,
Jeanette
--
* Website: http://juliencoder.de - for summer is a state of sound
* SoundCloud: https://soundcloud.com/jeanette_c
* Youtube: https://www.youtube.com/channel/UCMS4rfGrTwz8W7jhC1Jnv7g
* GitHub: https://github.com/jeanette-c
* Twitter: https://twitter.com/jeanette_c_s
I believe
We all have one true love
Somewhere in this world <3
(Britney Spears)
15 Mar
15 Mar
1:58 p.m.
On 14/03/19 19:41, Lorenzo Sutton wrote:> On 14/03/19 16:38, Mark
Raynsford wrote:
In theory this would be doable also in Pd (again with some minor
scripting glue and running it 'headless'). Problem is, I don't think
there's any external to load vst plugins in linux...
Well, looks like I stand corrected... Announcement about this came
through on the Pd mailing list earlier today :-)
https://git.iem.at/pd/vstplugin
Lorenzo.
7:32 p.m.
On 3/15/19 7:58 AM, Lorenzo Sutton wrote:
On 14/03/19 19:41, Lorenzo Sutton wrote:> On
14/03/19 16:38, Mark
Raynsford wrote:
"NOTE: currently only VST2.x plugins are supported but VST3 support
will come soon!"
"...get the Steinberg VST2 SDK and copy it into /vst..."
Hi, just a quick note: The VST 2.x SDK is no longer available,
and even the 'temporary' 2.x support within the 3.x SDK is gone
or will be gone soon.
Tim.
In theory this would be doable also in Pd (again with some minor
scripting glue and running it 'headless'). Problem is, I don't think
there's any external to load vst plugins in linux...
Well, looks like I stand corrected... Announcement about this came
through on the Pd mailing list earlier today :-)
https://git.iem.at/pd/vstplugin
Lorenzo.
14 Mar
14 Mar
8:41 p.m.
On 14/03/19 16:38, Mark Raynsford wrote:
Hello!
I'm attempting to produce a set of sound files that will be loaded into
a sampler. Given a MIDI file that contains a single channel containing
eight one-second notes, I'd like to be able to instantiate a VST plugin,
load a preset into it, pass the MIDI file through it, and record the
output to an audio file without any leading or trailing silence. I'd
like to be able to do that entirely from the command line without any
interactivity. I have a couple of hundred repetitions of this that I'd
want to perform, so as you can imagine I don't want to waste any time
on manual input.
Is there anything open-source out there that allows for this kind of
headless, command-line rendering?
Is the VST plugin and preset the same for all MIDI files? in that case
you could load the plugin and preset in e.g. Ardour as Jeanette suggested.
Then a scripted combination of commandline tools Jack-smf-utils [1],
jack_capture [2], jack_freewheel [3] might work.
In theory this would be doable also in Pd (again with some minor
scripting glue and running it 'headless'). Problem is, I don't think
there's any external to load vst plugins in linux...
Lorenzo.
[1] https://sourceforge.net/projects/jack-smf-utils/
[2] https://github.com/kmatheussen/jack_capture
[3] Included with JACK (I think)
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
https://lists.linuxaudio.org/listinfo/linux-audio-user
9:29 p.m.
I would check if synthclone can run headless. Synthclone is made for
cloning hardware synths as well as vst, lv2 and whatever.
On Thu, Mar 14, 2019, 19:41 Lorenzo Sutton <lorenzofsutton(a)gmail.com wrote:
On 14/03/19 16:38, Mark Raynsford wrote:
Hello!
I'm attempting to produce a set of sound files that will be loaded into
a sampler. Given a MIDI file that contains a single channel containing
eight one-second notes, I'd like to be able to instantiate a VST plugin,
load a preset into it, pass the MIDI file through it, and record the
output to an audio file without any leading or trailing silence. I'd
like to be able to do that entirely from the command line without any
interactivity. I have a couple of hundred repetitions of this that I'd
want to perform, so as you can imagine I don't want to waste any time
on manual input.
Is there anything open-source out there that allows for this kind of
headless, command-line rendering?
Is the VST plugin and preset the same for all MIDI files? in that case
you could load the plugin and preset in e.g. Ardour as Jeanette suggested.
Then a scripted combination of commandline tools Jack-smf-utils [1],
jack_capture [2], jack_freewheel [3] might work.
In theory this would be doable also in Pd (again with some minor
scripting glue and running it 'headless'). Problem is, I don't think
there's any external to load vst plugins in linux...
Lorenzo.
[1] https://sourceforge.net/projects/jack-smf-utils/
[2] https://github.com/kmatheussen/jack_capture
[3] Included with JACK (I think)
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
https://lists.linuxaudio.org/listinfo/linux-audio-user
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
https://lists.linuxaudio.org/listinfo/linux-audio-user
16 Mar
16 Mar
3:18 p.m.
Thanks, all who replied!
I'm still undecided as to which way I'm going yet. I should elaborate a
bit: I'm producing a SoundFont. Each SoundFont instrument is produced
by recording eight notes (C0, C1, C2, ...) into a single FLAC file
which is then sliced up programatically. Right now, I'm using Ardour
to export the notes, but this is painful and requires a lot of manual
labour.
I can make the following simplifying assumptions:
* The MIDI file used is always the same
* The VST plugin used is always the same (Dexed [0])
* The VST plugin program selection is always the same (program 0), but
the loaded bank will differ.
I think the "loaded bank will differ" part is probably the awkward bit:
I think loading a bank into Dexed is likely entirely plugin-specific.
I'm not sure if this step can be automated, and if it can't, that
pretty much kills the whole thing.
[0] https://asb2m10.github.io/dexed/
--
Mark Raynsford | http://www.io7m.com
2024
days inactive
2026
days old
linux-audio-user@lists.linuxaudio.org
6 comments
5 participants
participants (5)
-
Anders Hellquist -
Jeanette C. -
Lorenzo Sutton -
Mark Raynsford -
Tim