Hi!
Recently found
"Camilla-DSP" https://github.com/HEnquist/camilladsp
It's "a tool to create audio processing pipelines for applications such
as active crossovers or room correction."
I wanted to try out the room correction, but I have to create FIR
filters for my speakers. And this is where I got stuck.
My setup is as follows:
Ubuntu Studio 24.10
RME HDSP 9652 (output 1-8 = 4xHeadphone, output 9-17 connected with amp)
Dayton Audio 12-channel Amp (channel 9/10 not connected)
4 2-way-speakers (LF/HF), each chassis directly connected to the amp
(channel 1 - 8)
1 Subwoofer, directly connected to the amp (channel 11/12 bridged)
Tried to use REW, but I can't make it behave like an usual jack
application ("use input 1, output 9"), instead it deals with cryptic
sorround nomenclature and doesn't show up on qpwgraph…
Now, as I understood it right, my steps are:
* Sending a sweep to every speaker chassis, each, and record the output
(of each one) through a measurement microphone. (Measurement samplerate
= playback samplerate)
* "Invert" each file
* Make use of those files in "Camilla DSP"
Steps missing? Or am I totally wrong, here? What tools can I use?
Of course I have to limit the sweeps to frequencies and levels the
speaker chassis will survive…! (Especially for the Tweeters)
But one thing I don't understand is how the crossover frequencies are set…
Maybe someones able to turn the light on…
Thanks!
And greets!
Mitsch
Software/Hardware Hybrid:
1998
Yamaha DSP Factory DS2416 + AX16-AT
4-band eq,
Hi & Low bands can be switched to Hi-pass & Low-pass Filters *Unknown slope.
https://usa.yamaha.com/files/download/brochure/4/320264/ds2416.pdfhttp://www.oldschooldaw.com/manuals/cubase/CubaseSX/SX/PC-2002/cubaseSX_v1.…https://uk.yamaha.com/en/support/updates/index.html?c=proaudio&k=DS2416
Creamware Scope pci / Soniccore
for macOS 9.2 / PowerPC G4 G5
but also worked on Windows, and OSX 10.4
Ensoniq Paris ..
Software Only:
2006 - 2019
thuneau allocator for WinXP
https://web.archive.org/web/20190716065853/http://www.thuneau.com/allocator…https://web.archive.org/web/20140530212406im_/http://www.thuneau.com/images…
had -24dB/Oct, -42dB & -60dB/Oct
Hardware:
Yamaha D2040 "Digital" was interesting, F1040 "Analog"
JBL DSC 260 & 280 "Digital"
were rebranded BSS Omnidrive 260 & 280
JBL had Orange LCD display,
BSS had Blue,
maybe DACs were different JBL had 20-Bit,
dbx 260 "very popular" had different DSP algorithms,
included a digital version of the dbx 120xp / 120a sub-harmonic generator.
dbx has many models of the 260, 360, and very large Pro models 480 / 4800...
there were "rare" brands,
like Dynacord, EV, etc...
Analog:
DOD 835
dbx 234xl
JBL M553 / M552
Rane AC
Ashly XR / SC
Furman X-424 / X-324 / TX-3
Peavey
Crown
Crestb cpc 1234
Behringer
Symetrix 524e
were the "go to"
UREI 525 was more rare,
BSS FDS-318 was the most desired analog crossover from BSS.
all other Analog crossovers from BSS had Fixed resistor network cards, 318 was variable.
Klark Teknik DN800 was also fixed network cards, DN 8000 was "Digital"
QSC had an Optional Add-On Crossover Card for PLX, DCA & PL2 amplifiers, similar was included by default in PL2A
Peavey CS had "filter plug type crossovers" like vacuum tubes, but pre-made crossovers.
some speaker brands have its own crossover, like Yorkville, Apogee audio, Bose.
Crossovers could be done "Analog"
Horn loaded loudspeakers,
increasing the front chamber of the speaker before the Horn Throat.
creates a natural roll off in the highs.
Horns are "Band-Pass", and can be "Tuned".
Most Pro Audio Class-D amplifiers have DSP built-in, to replace external digital crossover units "lower cost".
Hardware outlast Software,
Software is Vaporware.
Hybrid depends,
sometimes 60/40 other times: 40/60.
for example:
RME hdsp9632 pci card
works flawless but...
Linux source code has small bugs since Ubuntu 13 probably Cosmic Ray Flip type,
bug accumulate, becoming almost unusable in 20.04 LTS with Alsa driver.
Most Analog crossovers & eqs emulate Coils / transformers using Op-amps + capacitors.
most have Servo-Balanced I/O's
to emulate Balanced I/O Transformers.
everything is an emulation of an emulation. LOL
aloop is a audio file looper for Linux using PortAudio as backend (jack,
pulse, alsa), libsndfile to load sound files and zita-resampler to
resample the files when needed. For varispeed, fine tuning and pitch
shifting it use librubberband. The GUI is created with libxputty.
This release add support for varispeed, fine tuning and pitch shifting,
contributed by @rubberplayer <https://github.com/rubberplayer>
<https://github.com/brummer10/aloop/blob/main/alooper.png?raw=true>
aloop comes with the following features:
* support all file formats supported by libsndfile.
* resample files on load to match session Sample Rate
* file loading by drag n' drop
* included file browser
* open file directly in a desktop file browser
* open file on command-line
* create, sort, save and load playlists
* select to loop over a single file or over the play list
* move play-head to mouse position in wave view
* set loop points for start/end loop
* save loop points in play list
* save selected loop as wav file
* play backwards
* volume control
* endless looping
* break playback (keyboard support space bar)
* reset play-head to start position (keyboard support courser left)
* varispeed
* fine tuning
* pitch shifting
To build from source please use aloop-v0.4.tar.xz as only this contain
the needed submodules to build aloop successful.
Dependencies
* libsndfile1-dev
* portaudio19-dev
* libcairo2-dev
* libx11-dev
* librubberband-dev
Project Page:
https://github.com/brummer10/aloop
Release Page:
https://github.com/brummer10/aloop/releases/tag/v0.4
This is the first release of NeuralRack.
NeuralRack is a Neural Model loader (based on Ratatouille), it allow to
load two model files and run them serial.
The input/output could be controlled separate for each model.
It supports *.nam files <https://tonehunt.org/models?tags%5B0%5D=nam>
with the Neural Amp Modeler
<https://github.com/sdatkinson/NeuralAmpModelerCore> engine, or *.json
or .aidax files <https://tonehunt.org/models?tags%5B0%5D=aida-x> with
the RTNeural <https://github.com/jatinchowdhury18/RTNeural> engine.
For tone sharping a 6 band EQ could be enabled.
NeuralRack provide a mono input and a stereo output.
Additional it allow to load up a separate Impulse Response file for each
output channel (stereo).
Neuralrack provide a buffered Mode which introduce a one frame latency
when enabled.
It moves the complete processing into a background thread. That could
reduce the CPU load when needed.
The resulting latency will be reported to the host so that it could be
compensated.
NeuralRack could be build as LV2 plug or/and as stand alone application.
Stand alone it could be used with JackAudioConnection Kit or with Alsa
(pipewire).
Project Page is here:
https://github.com/brummer10/NeuralRack
Release Page with pre-build binaries is here:
https://github.com/brummer10/NeuralRack/releases/tag/v0.1.1
Ratatouille is a Neural Model loader and mixer for Linux/Windows.
This release Fix a issue with the lv2 plugin from last 0.9.8 release
which leads to a crash on several hosts.
It also introduce a stand-alone version for Linux and windows.
Stand-alone app will auto save and reload the last used settings.
A prebuild binary of the app is provided here in
Ratatouille.lv2-app-v0.9.9-linux-x86_64.tar.xz/
Ratatouille.lv2-app-v0.9.9-win64.zip
Note that the binary is fully optimized, so when your CPU didn't support
x86-64-v3 the only way to get it is to compile it yourself.
Ratatouille allow to load up to two neural model files and mix there
output. Those models could be [*.nam files](https://tonehunt.org/all) or
[*.json or .aidax files](https://cloud.aida-x.cc/all). So you could
blend from clean to crunch for example, or, go wild and mix different
amp models, or mix a amp with a pedal simulation.
Ratatouille using parallel processing to process the second neural model
and the second IR-File to reduce the dsp load.
Ratatouille allow to compensate phasing issues between the loaded Models.
The "Delay" control could add a small delay to add some color/reverb to
the sound.
To round up the sound it allow to load up to two Impulse Response files
and mix there output as well. You could try the wildest combinations,
or, be conservative and load just your single preferred IR-File.
Each neural model may have a different expected Sample Rate, Ratatouille
will resample the buffer to match that.
Impulse Response Files will be resampled on the fly to match the session
Sample Rate.
Project Page (Source Code):
https://github.com/brummer10/Ratatouille.lv2
Release Page (Binaries):
https://github.com/brummer10/Ratatouille.lv2/releases/tag/v0.9.9
have fun
