Audiorealism ABL3

[Mike Janney]

In my personal opinion the 3.2.2 doesn't add much to 3.3.0, Model 2 is already very close to that engine except with improved filter response at higher cutoff + lower CPU consumption.
I'd be happy to take suggestions

One big reason to include 3.2.2 is for backward compatibility but only if you keep the plugin ID the same as 3.2.2, so it opens and recalls correctly in older DAW projects. The idea is, one ABL3 with all engines, not several ABL3 dll's with different engines, which would be a mess IMO.

We cannot keep the same plugin ID because we had to drop 32-bit in order to build M1 from the case code base without endless hassle. The session data saved with the host and not really compatible between the versions.

[midi_transmission]

Glad that you took no offence! It was also not meant in a bad way, as said, I hope that it's rather perceived as a compliment at the same time that I think of 3.2.2 VCO2 as almost perfect.

Maybe the difference is not the VCO, I can't say what the difference is technically. But one example, I notice that 3.2.2 and 3.3.0 have a different transient behavior, 3.3.0 Model2/VCO2 has sharper transients than 3.2.2 VCO2. Some may prefer it, I like the silky character of 3.2.2.

By the way, any chance that you may approach a 909 emulation with the same attention to detail? D16 is ok, but has some weak points and there seems no interest in updating it since a couple of years. Roland is out of the game with their cloud craziness. And samples are static or difficult to tweak.

I like that ABL is not just a sound module but works also great on a computer screen ui wise - even easier to use than a real 303. This is something what D16 is missing too with their 909, I think roland has some ideas in this direction, but it's out of the game due to their cloud system.

[JamminFool]

By the way, any chance that you may approach a 909 emulation with the same attention to detail?

do you mean that you are wondering if there will be an update to ADM (Audiorealism version of the 909/808/606)?

[Mike Janney]

Spent some mode time checking my notes and the emulation models. Here's what I've gathered:
- Model 1 is pretty much ABL2's sound engine (late one). This one does not perform that well at 96kHz.
- Model 2 is very similar to ABL3.2.2's model except it has an improved filter, lower CPU consumption. This one performs well at 96kHZ and is designed for that.
- Model 3 is the one closest to the lab 303 (a good condition one) and the one we use for shoot outs. Its design to sound like a 303 pretty much out of the box without any extra EQ or stuff like that. This one performs well at 96kHZ but taxes the cpu a bit more.

In a test build here I put the exact 3.2.2 engine back and switched back and forth between Model 2 and that. Except that Model 2 is brighter, and some of the parameters are scaled differently they are very similar. Model 2 IS a better version of 3.2.2.

Model 3 has lots of changes over 3.2.2 over the years and was developed with the ear of some high profile 303 enthusiasts. But even after all that work some preferred Model 2 and even Model 1 in some tests.

This is work that went on over some years so there is some confusion even here why and where all the changes are made. I will attempt to demonstrate the different models compared to our lab 303 and/or some other clones in a stream or something.

I like the suggestion of naming them like this:
ABL2
ABL3.2.2
ABL3.3 (Model 1)
ABL3.3 (Model 2)
--Mike

[Mike Janney]

By the way, any chance that you may approach a 909 emulation with the same attention to detail? D16 is ok, but has some weak points and there seems no interest in updating it since a couple of years. Roland is out of the game with their cloud craziness. And samples are static or difficult to tweak.

Yes, I have been working on the 909 and 808 emulations.. I rewrote the models for the kick and snare for the 909. We do have something to roll out in that regard, just hard to say when/how. We still have some work to do on the 808/909 clap.

Static samples will never be able to capture the nuances of the 808/909, though a sample is better than a bad model in any case.. We use hybrid approaches where possible, but only if the results are good.

[Mike Janney]

-- oops double post

[JamminFool]

Yes, I have been working on the 909 and 808 emulations..

this is great to hear Mike! looking forward to it.

(maybe ABLPro will get some love too. )

[midi_transmission]

I like the suggestion of naming them like this:
ABL2
ABL3.2.2
ABL3.3 (Model 1)
ABL3.3 (Model 2)
--Mike

Makes perfectly sense to me. Really like that solution. Kudos for the direct exchange and constant polishing in general.

The details about the models are helpful, thanks for that. I think the technically worse filter maybe a part why I like 3.2.2? It's sounds softer, less hifi so to say.

Looking very much forward to the 808/909 emulation. Fully agree with your thoughts about samples. Even when they are multi-sampled. It's ok, but it's not the same. The sequencer design, interaction of sounds played at the same time and the living, non-static behaviour of the triggered sounds are a big part of the appeal.

[gentleclockdivider]

Spent some mode time checking my notes and the emulation models. Here's what I've gathered:
- Model 1 is pretty much ABL2's sound engine (late one). This one does not perform that well at 96kHz.
- Model 2 is very similar to ABL3.2.2's model except it has an improved filter, lower CPU consumption. This one performs well at 96kHZ and is designed for that.
- Model 3 is the one closest to the lab 303 (a good condition one) and the one we use for shoot outs. Its design to sound like a 303 pretty much out of the box without any extra EQ or stuff like that. This one performs well at 96kHZ but taxes the cpu a bit more.

In a test build here I put the exact 3.2.2 engine back and switched back and forth between Model 2 and that. Except that Model 2 is brighter, and some of the parameters are scaled differently they are very similar. Model 2 IS a better version of 3.2.2.

Model 3 has lots of changes over 3.2.2 over the years and was developed with the ear of some high profile 303 enthusiasts. But even after all that work some preferred Model 2 and even Model 1 in some tests.

This is work that went on over some years so there is some confusion even here why and where all the changes are made. I will attempt to demonstrate the different models compared to our lab 303 and/or some other clones in a stream or something.

I like the suggestion of naming them like this:
ABL2
ABL3.2.2
ABL3.3 (Model 1)
ABL3.3 (Model 2)
--Mike

I post again for you to re-consider the labbeling of the treble knob
When in model 2 advanced mode on , setting treble to 0 % removes highs ,leaving it at the (default) middle position of 50% it sounds almost like the 3.2.2 mode
So it makes sense that the knob is bipolar , and the midlle position =0 , counterclockwise = negative , clockwise = positive

[Echoes in the Attic]

Spent some mode time checking my notes and the emulation models. Here's what I've gathered:
- Model 1 is pretty much ABL2's sound engine (late one). This one does not perform that well at 96kHz.
- Model 2 is very similar to ABL3.2.2's model except it has an improved filter, lower CPU consumption. This one performs well at 96kHZ and is designed for that.
- Model 3 is the one closest to the lab 303 (a good condition one) and the one we use for shoot outs. Its design to sound like a 303 pretty much out of the box without any extra EQ or stuff like that. This one performs well at 96kHZ but taxes the cpu a bit more.

In a test build here I put the exact 3.2.2 engine back and switched back and forth between Model 2 and that. Except that Model 2 is brighter, and some of the parameters are scaled differently they are very similar. Model 2 IS a better version of 3.2.2.

Model 3 has lots of changes over 3.2.2 over the years and was developed with the ear of some high profile 303 enthusiasts. But even after all that work some preferred Model 2 and even Model 1 in some tests.

This is work that went on over some years so there is some confusion even here why and where all the changes are made. I will attempt to demonstrate the different models compared to our lab 303 and/or some other clones in a stream or something.

I like the suggestion of naming them like this:
ABL2
ABL3.2.2
ABL3.3 (Model 1)
ABL3.3 (Model 2)
--Mike

I post again for you to re-consider the labbeling of the treble knob
When in model 2 advanced mode on , setting treble to 0 % removes highs ,leaving it at the (default) middle position of 50% it sounds almost like the 3.2.2 mode
So it makes sense that the knob is bipolar , and the midlle position =0 , counterclockwise = negative , clockwise = positive

If it's cutting treble when left of middle, then it should definitely be 0 in the middle position and negative to the left of it.

Is the bass not the same? It only adds bass?

[Mike Janney]

Hi Guys,

Uploaded an update with some of suggestions from this thread, including the mythical 3.2.2 audio engine.

- Fixed display of Parameters (In Live etc). Tuning now says (+/- semitones) and other parameters such as Cutoff 0.00 - 1.00
- Adding Model 4 (which is ABL3.2.2 due to request from KVR users)
- Changes to Emulation Engine menu entries to:
"ABL2"
"ABL3.2"
"ABL3.3 (Model 1)"
"ABL3.3 (Model 2)"
- Bass/Treble boost now reads +/- 0-6dB instead of 0-100%. But only in the gui, the parameter display from host is still 0.0-1.0

More info here :
https://www.audiorealism.se/news/audior ... -line-3301

[Funkybot's Evil Twin]

including the mythical 3.2.2 audio engine.

Thanks Mike!

[gentleclockdivider]

Hi Guys,

Uploaded an update with some of suggestions from this thread, including the mythical 3.2.2 audio engine.

- Fixed display of Parameters (In Live etc). Tuning now says (+/- semitones) and other parameters such as Cutoff 0.00 - 1.00
- Adding Model 4 (which is ABL3.2.2 due to request from KVR users)
- Changes to Emulation Engine menu entries to:
"ABL2"
"ABL3.2"
"ABL3.3 (Model 1)"
"ABL3.3 (Model 2)"
- Bass/Treble boost now reads +/- 0-6dB instead of 0-100%. But only in the gui, the parameter display from host is still 0.0-1.0

More info here :
https://www.audiorealism.se/news/audior ... -line-3301

There is just one problem , you included the 3.2.2 engine without improved emulation mode ( it's not working in 3.2.2 Mode ) , and this is what made 3.2.2 so good and a much equestd feature
VCF TRim is also not working in 3.2.2 mode
- The cutoff scaling is also OFF , 3.2.2 filter freq . goes much lower compared to 3.2.2 model in 3.x because improved emu switch is not working in 3.x ( 3.2.2 mode )
An lastly 3.2.2 mode uses a 50 % more cpu compared to the o.g. 3.2.2 , 10% versus 15 % ( it's now th mos cpu intensive model )
I think you included the 3.2 model and not the 3.2.2 ,since that version also didn't have the improv.emu switch and vcf trim
Sorry , still NO abl to rule them all

[gentleclockdivider]

Here are three notes played by keyboard , everything set to zero except env amount +decay at full
First three notes are abl 3.2.2 ( improved emulation off sounds exactly like 3.x in 3.2.2 mode ) , then 3.2.2 improved emu on (filter frequency is now much lower ) , then 3.x ( 3.2.2 mode .)
https://app.box.com/s/kw8ajvcbhsqmpixpumw35d90friwb38n

[gentleclockdivider]

Also encounterd a bug , switching to abl 2 resulted in silence
WIll try to replacte the bug