Hi!
Don't know, if this is actually the academic or scientifically
correct way to do it, but I got good sounding results - in fact, I
managed to insert mono-recorded voices into an ambisonic recording
with other people and it didn't sound that weird…
But I'm always interested in getting things more and more perfect, so I hope many of the ambisonics-pros are joining this thread to discuss the topic.
This is the software and hardware I used:
- Røde NT-SF1 1st order Ambisonics microphone
- Sound Devices MixPre-6 recording device
- Ubuntu Studio 04.22
- Carla 2.4.2
- Røde Soundfield PlugIn (Windows 64bit VST3)
- Wine
- Ardour 6.9
- Audacity 2.4.2
- LSP-PlugIn Suite (LV2)
- IEM Ambisonics PlugIn Suite (LinuxVST)
And here is the whole procedure, how I made it…
I made an ambisonics recording with people in a room. After that
I took a paper bag to create an impulse, about 50cm away from the
NT-SF1. (Just clapping the hand was too silent in my opinion and I
didn't want to use balloons because of the plastic waste. Paper
bags seem to be an relatively environment friendly and cheap
alternative to me.) I retried as often as it needed to get a good
sounding and non distorted record of the impulse and its respond.
Getting back to home I copied the A-format of my recordings to my
computer, fired up Ardour, created a 4-channel track and inserted
the 16-channel-version of carla patchbay into that channel strip.
I had to deactivate the panner in every channel strip to make
Ardour "ambisonics compatible", yes, also in the "Master"-channel
which must be blown up to 4 channels, too. I imported the tracks
that I wanted to be converted.
The only tool in my procedure that wasn't free and open was Rødes
Soundfield PlugIn, so I had to use some tricks to get manufacturer
proven A-to-B conversion of my recordings. FalkTX did make carla
windows-VST-capable, but it's not a bullet prove task… So, I got
the Soundfield exe-file, installed it with wine and I finally got
a dll I could drag and drop into carla patchbay's GUI. There it
is! Chances are Ardour crashes, so I made sure to save after every
step. As input format, I have to chose "NT-SF1", on output the
best choice is to use "B-Format (Ambix)", because Ambix is the
actual standard and the IEM plugins only handles Ambix. Of course,
I had to connect Carla's output to the input of Soundfield and
vice versa.
Time for the A-to-B-convertion. Drag and drop the 4-channel-audio
into the editor window of Ardour, set "Start" and "End" and use
export. Make sure to have all 4 channels to be exported into one
file. You can use both WAV or FLAC. Next step is Audacity. (You
don't need to close Ardour, sooner or later it will crash… :) )
Instead of stereo files, Audacity handles 4-channel-audio as 4 seperate mono files. (But it can be exported to 4-channel-audio, again, if you choose "advanced mix options" in the "import/export" division of the preferences. You won't need it for the virtual ambisonics IR reverb.) But - it is mandatory to keep the audio sample accurate between all the tracks! If you want to delete, delete from all of them, simultaneously! So, you choose your favorite BANG! out of the inpulse series you recorded and delete the rest. Export every mono track as WAV. Use Ambix nomenclature: 1st track is W, 2nd is Y, 3rd is Z and 4th is X. Best bet is to write the number and the letter in the name of the file. I choose: [nameoftheplace]_1_(W).wav and so on…
Now, to actually build the virtual ambisonics IR reverb, fire up Ardour and make it ambisonics compatible (deactivate panner). Create a 4-channel-audio-bus and place a 16-channel Carla patchbay. In the Carla patchbay, create 4 "LSP Impulse Responses Mono" - one for each channel. (Using LV2 in Carla works pretty stable!) Open the GUI for each "LSP Impulse Responses Mono" and use the corresponding WAV in it by clicking into the GUI and choose the right file. Inside Carla patchbay it should look as follows:
Carla channel 1 output ---> LSP Impulse Responses Mono #1
(with [nameoftheplace]_1_(W).wav) ---> Carla channel 1 input
Carla channel 2 output ---> LSP Impulse Responses Mono #2 (with
[nameoftheplace]_2_(Y).wav) ---> Carla channel 2 input
Carla channel 3 output ---> LSP Impulse Responses Mono #3 (with
[nameoftheplace]_3_(Z).wav) ---> Carla channel 3 input
Carla channel 4 output ---> LSP Impulse Responses Mono #4 (with
[nameoftheplace]_4_(X).wav) ---> Carla channel 4 input
Make sure, that every LSP IR Mono instance uses exactly the same values of headcut, tailcut and amplification. WIth headcut, you should cut the impulse. Now, you can use a mono audio file, place it into the ambisonics sphere f.e. with the IEM StereoEncoder, use an auxiliary send (don't forget to disable the panner of it, too!) and enjoy the reverberation! (Play with Ardour's "strict E/A" to get as much channels for each strip you need!) Using an EQ after the virtual reverb is always a good idea. LSP IR Mono has one integrated, but my impulses seem to have no linear frequency band and needed more than the +/-12 dB the plugin offers… (I take IEMs "MultiEQ" for this task - one setting for all the channels is another advantage.)
Unfortunately, Carla can't recall LSP IR Mono's WAV-file, so, you have to reload it, every time you start your session.
Greets!
I hope, this can be useful for anyone…
Mitsch
_______________________________________________ Linux-audio-user mailing list Linux-audio-user@xxxxxxxxxxxxxxxxxxxx https://lists.linuxaudio.org/listinfo/linux-audio-user
