What's new

Aaron Venture Infinite Brass - Update v1.2 is live!

HelixK

Active Member
Congratulations on the release Aaron. How exactly does this work? Is this sort of like Sample Modeling? I'm a little confused how you can have 3 mic positions for each of 60 seating options??
Great question, I've read the release notes a few times and failed to grasp how this is going to work. Built-in spatialization and positioning?
 

axb312

Senior Member
That was sudden and unexpected!

Well done though.

Personally dislike the font on the website and would like to know where this was recorded and how much disk space it takes up (disk space is becoming a precious commodity nowadays).

Nice that you're being upfront about the price as well and you included the MIDI files!
 
OP
aaronventure

aaronventure

Senior Member
Congratulations on the release Aaron. How exactly does this work? Is this sort of like Sample Modeling? I'm a little confused how you can have 3 mic positions for each of 60 seating options??
Thanks Noam!

I'm don't know how Sample Modeling does what they do, but the source samples in Infinite Brass don't really sound pretty. They're perfectly phase-aligned so you will always perceive it as a single voice, even though it's always flying between two layers (pp and mp, mp and mf, or mf and ff) which is why it's so sensitive and expressive.

Infinite Brass was recorded dry, in-the-bell, in a small studio. The positions and space are done through convolution. There are over 2600 recorded and edited impulse responses that I created specifically for Infinite Brass.

how much disk space it takes up
The size on disk is 6.95 GB. I'll update the post as well, thanks.
 

axb312

Senior Member
Thanks Noam!

I'm don't know how Sample Modeling does what they do, but the source samples in Infinite Brass don't really sound pretty. They're perfectly phase-aligned so you will always perceive it as a single voice, even though it's always flying between two layers (pp and mp, mp and mf, or mf and ff) which is why it's so sensitive and expressive.

Infinite Brass was recorded dry, in-the-bell, in a small studio. The positions and space are done through convolution. There are over 2600 recorded and edited impulse responses that I created specifically for Infinite Brass.


The size on disk is 6.95 GB. I'll update the post as well, thanks.
Thank you. Other questions :
- how many recorded dynamic layers and round robins ?
- list of articulations ?
- Were legato transitions recorded or....?
- were mutes recorded or are they simulated?
 
Last edited:
OP
aaronventure

aaronventure

Senior Member
Thank you. Be nice if you could put in how many recorded dynamic layers and round robins it has as well...
Recorded dynamic layers are 4, but that's not important because they are phase aligned and you are always somewhere between two layers. Never, ever will you only listen to a single dynamic layer for longer than a few moments, and that's only if your CC1 is 127 or 0. Standard sampled libraries that try and tackle solo instruments have really small crossfades in order to avoid phasing. You are mostly just controlling volume, and then have rough jumps between layers.

In Infinite Brass, you cannot perceive more than 1 voice per instrument, since the layers are perfectly phase-aligned, and the only thing the crossfade is doing is fading in/out the differences between the layers—overtones. And since the crossfades are as long as the layers themselves, it all actually feels like a crescendo/decrescendo of a real instrument.

round robins it has
For the most part, the instruments aren't linear. Pitch fluctuations are handled (amplified on top of the recorded ones/whatever is left after auto-tuning) by the script, and they're randomized. Breath fluctuation is randomized. There are multiple parameters regarding note starts being randomized every time you play a note. Usually it's 127 different possible outcomes for most parameters, but in some cases (like the pitch envelopes during the attack) it's 127x127. Legato has 127 different durations, depending on how hard you hit the note.

You play trills by just trilling on the keyboard—hold the starting note and repeat the landing note (see Steel Ballroom screencast), and it's almost impossible for you to play a trill line the exact same way twice. No chance that you hit the same CC1 curve, note timings, note durations and note velocities twice. So as long as you don't copy and paste the trill to all the other instruments, these will be the most natural-sounding trills you've written. Or rips, falls, diverging clusters, bends... You just play it all in.

The clams aren't scripted in; the random detunes, slippery fingers, etc. I'm against shoving that kind of stuff into a user's face, so you can do the clams yourself. Clam while you're playing, or include a short "wrong" note before or after a note. Use pitch bend to dial in detunes.

None of it will sound wrong, because the convolution comes after all of that. You can play 10 notes with same velocities, but even then they'll have different pitch envelopes, embouchure resonance envelopes and note durations, which will ultimately, coupled with convolution, sound like a series of 10 different notes.

Add in CC1 and different velocities and the possibilities are... infinite.

Or somewhere in the ballpark.
 

Nicola74

Member
Wow!! Looking forward for walkthrough...
Interesting choice: 4 trumpets and 5 trumpets with mute, 4 french horns and 2 french horns with mute...
Anyway It sounds great!!

Edit: made a mistake, I think you mean 5 kind of mutes for trumpets and so on...wow!
 
Last edited:

zolhof

Active Member
2018 shall be known as the year of the brass!

Congratulations, Aaron and Alexander, looking forward to the walkthrough. Great sound and such a small footprint.

Also, nice selection of demos, I can feel the love in these tracks. :)
 

HelixK

Active Member
Aaron, could you share which microphones were used and what was the recording process like? Signal chain and such.
 
Top Bottom