×
What's new
By:

Samplicity announces Berlin Studio plugin

Kalli said:
This is fascinating! I learned a lot just from reading this post. Please don't remove the delay options! I didn't fully understand their purpose at first, but will start experimenting now.

UPDATE: Whoa! Increasing the delay of the dry signal really helps glue the signals together. Thanks for the tip!

Case in point: CSW solo flute (close mic only) sounds much more focused with the correct delay setting (18 ms):

Delay OFF:

View attachment CSW Flute 0 ms.mp3

Delay 18 ms:

View attachment CSW Flute 18 ms.mp3

Dry (for reference):

View attachment CSW Flute Dry.mp3


@Peter Emanuel Roos Is there any risk that slightly wrong delay values will introduce phasing issues?

Do you happen to know what the delay of the dry signal should be for the different string sections?
Click to expand...
In a free, not so long away update there will be an auto align option. Not by delaying the dry input, but by skipping the corresponding part in the three mic sets (with fades etc, of course). I will use detailed distance information that I have, not measured in situ, but reconstructed from delay information / speed of sound, etc etc
Maybe half december? I have used it myself, and find it pretty well working.
I happened to check the woodwinds today: 14 msec, and violins 1 7 msec. Closest L or R distance to the Decca center is used.
And on the Decca: maybe in a later version I will offer 3 mono channels Decca L, C and R. As an option. Maybe too confusing for some.
 
tabulius said:
It is not uncommon to align close signals to room mics by mixing engineers, so it is great that this is build in. Actually, it would be great that many sample developers would do the same with their room mics.

The cpu usage is impressive, but I’m a bit worried how my future AMD pc will perform with this plugin, but I hope it will run it well enough.
Click to expand...
Higher up, a few days ago I replied with a long text to Hendrik Buus Jensen about this time alignment feature, that will be included in an end-of-year update.

The efficiency of Berlin Studio, and other Samplicity plugins will only improve, because the current version has its multi-threading switched off. This is related to reported issues with the Juce library (that I use) and threading on Silicon chips. Berlin works fine (3 true stereo convolutions in paralel) in the host's callback, but that was not the meant to be the final approach.

The plugins use SIMD libraries that take advantage of the user's processor feature, so if you have AV512 the plugin will use that and be more efficient. The overall compiled code "only" uses AVX optimization, as I know that some people are still using 10 year old computers. I am considering to offer fully AVX512 (and Neon optimized) versions on special request.

Finally, there will be "lossy" convolution options, that you cannot tell apart from the "exact" convolution. Expect 25-50 % less CPU usage. These are not future ideas, but features that passed their Proof-of-concept phase.
 
Peter Emanuel Roos said:
In a free, not so long away update there will be an auto align option. Not by delaying the dry input, but by skipping the corresponding part in the three mic sets (with fades etc, of course). I will use detailed distance information that I have, not measured in situ, but reconstructed from delay information / speed of sound, etc etc
Maybe half december? I have used it myself, and find it pretty well working.
I happened to check the woodwinds today: 14 msec, and violins 1 7 msec. Closest L or R distance to the Decca center is used.
And on the Decca: maybe in a later version I will offer 3 mono channels Decca L, C and R. As an option. Maybe too confusing for some.
Click to expand...
No, not at all confusing. Please add in the separate mono mic channels. :)
 
Peter Emanuel Roos said:
In a free, not so long away update there will be an auto align option. Not by delaying the dry input, but by skipping the corresponding part in the three mic sets (with fades etc, of course). I will use detailed distance information that I have, not measured in situ, but reconstructed from delay information / speed of sound, etc etc
Maybe half december? I have used it myself, and find it pretty well working.
I happened to check the woodwinds today: 14 msec, and violins 1 7 msec. Closest L or R distance to the Decca center is used.
And on the Decca: maybe in a later version I will offer 3 mono channels Decca L, C and R. As an option. Maybe too confusing for some.
Click to expand...
@Peter Emanuel Roos , congratulations on what appears to be quite a superb accomplishment! The auto align feature sounds very interesting. Looking forward to hearing some comparative examples (with the feature on and off). Do I remember reading that there was a slight issue with apple silicon at the moment? Perhaps I'm mistaken and this has already been addressed. Look forward to picking this up soon. Very impressive work.
 
NeonMediaKJT said:
that'd be great
Click to expand...
Here's a comparison of Mozarteum and Berlin Studio using one of the @aaronventure demos to minimize my ability to butcher a comparison to just mixing. :)

For the Berlin Studio version, I have all AV instruments set to Studio IR with close mics solo'd and maxed in volume.

AV Steel Ballroom Mozarteum (Mix Mic 3 on all instruments):
View attachment AV Steel Ballroom (Mozarteum).mp3

AV Steel Ballroom Berlin Studios (-6 dB direct, 0 dB Decca, -9 dB AB, -12 dB Surround):
View attachment AV Steel Ballroom (Berlin Studio).mp3
 
Trash Panda said:
下面是使用 @aaronventure 演示之一对莫扎特音乐和柏林工作室进行的比较，以尽量减少我将比较的能力与仅混音的比较。:)

对于 Berlin Studio 版本，我将所有 AV 乐器设置为 Studio IR，并使用近距离麦克风独奏并最大化音量。

AV Steel Ballroom Mozarteum（在所有乐器上混合 Mic 3）：
View attachment AV Steel Ballroom (Mozarteum).mp3

AV Steel Ballroom Berlin Studios（-6 dB 直接，0 dB Decca，-9 dB AB，-12 dB 环绕声）：
View attachment AV Steel Ballroom (Berlin Studio).mp3
Click to expand...
Did you use any panning?
 
And from a PC usage perspective, on my Surface Book 2 laptop (i7-8650U with 16 GB RAM), I was able to run 29 instances of AV Infinite Winds/Brass through 5 instances of Berlin Studio (Woodwinds, Trumpets, Horns, Trombones and Tuba) without overloading the CPU or RAM.
 
Peter Emanuel Roos said:
In a free, not so long away update there will be an auto align option. Not by delaying the dry input, but by skipping the corresponding part in the three mic sets (with fades etc, of course). I will use detailed distance information that I have, not measured in situ, but reconstructed from delay information / speed of sound, etc etc
Maybe half december? I have used it myself, and find it pretty well working.
I happened to check the woodwinds today: 14 msec, and violins 1 7 msec. Closest L or R distance to the Decca center is used.
And on the Decca: maybe in a later version I will offer 3 mono channels Decca L, C and R. As an option. Maybe too confusing for some.
Click to expand...
Thank you, Peter! An auto align option does sound like a great idea. Looking forward to that update.

And thanks for checking the delay times! Experimenting had led me to ~8 ms for violin 1, so I'm glad to hear that I was in the ballpark. :)

I don't think the three Decca channels would necessarily be too confusing. But it would probably be a good idea to have their settings (gain etc.) linked by default, with an option to unlink for more advanced tweaking. I suspect that for most use cases, it will be enough to use it in linked mode. Also, the GUI may become a little cluttered with three separate Decca channels, so perhaps it would be a good idea to show the Decca tree in a collapsed state as default, with the option to expand it into three channels for advanced tweaking. I've no idea how easy/difficult that would be to implement, so it's just a thought. I like the relatively clean GUI in the current version, though.
 
Peter Emanuel Roos said:
This is a known issue now, and all focus is on fixing this. In the mean time, users have reported that the plugin works well if you activate it yourself in the PluginManager.
Click to expand...
Even with manual activation in Logic's PluginManager the plugin doesn't show up in the plugin list here.
I'm on a M1 MacStudioUltra, Monterey 12.6.1
I tried Logic 10.6.3 and Logic 10.7.5
 
Peter Emanuel Roos said:
Higher up, a few days ago I replied with a long text to Hendrik Buus Jensen about this time alignment feature, that will be included in an end-of-year update.

The efficiency of Berlin Studio, and other Samplicity plugins will only improve, because the current version has its multi-threading switched off. This is related to reported issues with the Juce library (that I use) and threading on Silicon chips. Berlin works fine (3 true stereo convolutions in paralel) in the host's callback, but that was not the meant to be the final approach.

The plugins use SIMD libraries that take advantage of the user's processor feature, so if you have AV512 the plugin will use that and be more efficient. The overall compiled code "only" uses AVX optimization, as I know that some people are still using 10 year old computers. I am considering to offer fully AVX512 (and Neon optimized) versions on special request.

Finally, there will be "lossy" convolution options, that you cannot tell apart from the "exact" convolution. Expect 25-50 % less CPU usage. These are not future ideas, but features that passed their Proof-of-concept phase.
Click to expand...
Berlin Studio seems to run flawless on my 2010 MacPro slave with High Sierra (as AU within VE-Pro). Only inside Logic on my iMac with BigSur the CPU overloads immediately when launching the 1st instance. I impatiently expect the update o_O
 
This does sound excellent to me. One request for a future update would be scaleable UI, or perhaps a mixer that can collapse as the UI is pretty large on my MacBook screen, so a smaller UI would help with the screen real estate.

It does sound great though and makes blending quick and easy.
 
Wow, Peter, this is really amazing - you've come a long way since your first dealings with IR's!:) Congratulations - it must have been a monumental task to create this plugin!
 
You must log in or register to reply here.
Back
Top Bottom