Check out Audio Impressions’ Realtime Instruments

[dwdonehoo]

Audio Impressions’ Realtime Instruments: I have been looking into this and I am impressed. Go here and watch the whole movie:
http://www.audioimpressions.com/ (http://www.audioimpressions.com/)
Real time articulation changes, distributed processing, real humanizing, divisi sizes do not change just because you play different numbers of multiple notes, realistic glide, variable room size (multiple variable mikes), a massive number of instruments available, delivered on hard drives, 3rd party library integration and more. These guys are really going for the orchestral realism holy grail. Even with the funky video sound, it sounded very good. I would like to hear some demos when available.
Gary Garritan has commented on this product in his NAMM report. It may be a bit pricey depending on the depth of your implementation. The website video may be the most recent demo they have at this time.
Comments?

[alx]

It certainly has some good ideas and features implemented. Truth be known I was interested in this until I found out about the system requirements.

In excess of 5 top of the line dedicated computers are required to use the majority of the orchestra. If you just want strings, you need at least 2 dedicated machines, possibly 3.

The sound on the video isn't good enough to make a sound judgement really, it could still go either way in that regard in my opinion.

From what I've heard there's still no real sounding scripted legato either.

Still, we shall see in time...

[alprazolam]

...video is too old and sonic quality is not good. Waiting for a demo justifing
pricing and amount of harware-software.

[Marcussen]

Damn.. I thought it was a new movie! :D

[dpasdernick]

This library has sure got me curious. I really like the thought process behind it and, IMHO, it sounds really nice. It also seems really playable. However, how can you get all of the strings recorded with all of their articullations and velocities in an under 5 gigabyte footprint? Sounds too good to be true. And if you can do it, why do you need so many computers to toss around 5 gigs of data?

VSL makes some outstanding software instruments. Their solo strings take 82 gigabytes of disk space. They have sampled the daylights out of the solo strings and it shows, both sonically and file size wise. How do you get this type of great sound without going for the "brute force/huge filesize approach?

Next look at all of the instruments that this library will have including stuff like guitars, drums, world stuff, big band stuff. 600 sampled instruments in all! What an ambitious project! Hope it's all that and a bag of chips!

Now I just need to find the 20-30k and keep it secret from the wife!

2 cents...

Darren

[dwdonehoo]

Damn.. I thought it was a new movie! :D

Well, OK! :) Here is a link to a movie from Winter NAMM that the Audio Impressions guys sent me:

http://keyboardmag.tv/index.html?req=1&station=kbnamm07&video=kbtv/kbnamm07/kbtv_namm07_audioimpressions

Like I and Darth Vader said, "Impressive...very impressive.":hp:

[Chris Hein]

Wow,
they got a Mac award for a pc based product that has not been released yet.
Again, impressive!

Chris Hein

[ed hamilton]

They have to stop relying on these low quality videos for promotion.
DEMOS please!!!!
It's two full years now and all the buzz is based on two super low bandwidth crappy video sound. And it seems those that played it in person at namm are far from glowing about it.

Waiting for high quality demos.

[Ben H]

It does sound better than the old demo.

[Marcussen]

Well, OK! :) Here is a link to a movie from Winter NAMM that the Audio Impressions guys sent me:

http://keyboardmag.tv/index.html?req=1&station=kbnamm07&video=kbtv/kbnamm07/kbtv_namm07_audioimpressions

Like I and Darth Vader said, "Impressive...very impressive.":hp:

Thanks for that :)

Now we just need demos.

[JMDNYC]

I've read all the online documentation and I've decided that, yes, I would like to buy it. However, I would never want to buy six new p.c. (I'm a Mac guy), and there are still no indications of price nor any demos.

I'm hoping in the years to come they'll rework some of the patented technology and sell it in LE and cross-platform versions. Seeing how quickly hardware moves, in three years we'll probably be able to run it all on three mac mini's (which will all be the size of the clip-on ipod).

A lot of the concept seems similar to what Gary Garritan is doing, but Gary's doing it with a pocketbook-friendly approach.

[alx]

Well, OK! :) Here is a link to a movie from Winter NAMM that the Audio Impressions guys sent me:

http://keyboardmag.tv/index.html?req=1&station=kbnamm07&video=kbtv/kbnamm07/kbtv_namm07_audioimpressions

Like I and Darth Vader said, "Impressive...very impressive.":hp:


Thanks Doyle, that was interesting, if only for the reason that I got to hear something new about this bloody thing!

Marcussen - I can tell you're as excited as I am for new info :) The long drawn out hype is getting old...

I'm still waiting for something that demonstrates scripted legato. Something like the Garritan solo violin scripts applied to each desk and solo instrument would sound awesome, especially when combined with the random time / pitch spread.

[Will Roget]

I'm not sure if I've mentioned this before, but what excites me most about DVZ is how it handles dissonant / clustered / atonal passages. There's a natural tendency to detune a bit when playing a note within a very dissonant chord, and often the timing appears more scattered than usual within a section as well. Traditional libs don't have access to this kind of random pitch spread within a section, and so tons of dissonance effects are impossible unless they've been sampled specifically (which, of course, limits their versatility).

[geronimo001]

They have to stop relying on these low quality videos for promotion.
DEMOS please!!!!
It's two full years now and all the buzz is based on two super low bandwidth crappy video sound. And it seems those that played it in person at namm are far from glowing about it.

Waiting for high quality demos.

Yea! DEMOS please!

[dwdonehoo]

Yea! DEMOS please!

I really don’t think demos, or the sound, is the important thing here. For one thing they are promising integration with all the major libraries out there (ReMap), and you already know what they sound like. Even if their own instruments sound like crap (and I bet they won’t), that is not the important thing either.
The important thing here is “The Idea”. In this case, “The Idea” is the solving of many of the remaining problems of dealing with multi-articulated sample based instruments, and delivering a streamlined workflow from musical idea to finished composition with a great reduction of the usual effort involved. This promises to be a unified platform package where many long desired features are under one roof: real time articulation changes, distributed processing, real humanizing, divisi sizes (number of instruments playing at once) does not change no matter mow many notes are being played, realistic glide, variable room size with player positioning, real-time pitch manipulation, forced transposition, scoring program integration and more. After I started this thread, Audio Impressions called me and they say the stuff they are talking about now is just the tip of the iceberg, and there is a huge amount of tech waiting in the wings. Stay tuned…

Check out their website:
http://www.audioimpressions.com (http://www.audioimpressions.com/)

[geronimo001]

Well it does sound interesting when you put it that way... Sounds like a BIG project thou. I'm no expert gut it sounds to me like a lot of things could go wrong, i dunno?.. But they sure deserve my respect for this huge undertaking, and i wish them all the best! :)

:|:

[JohnCarter]

I watched the 07 namm video about DVZ and despite it sounds like a good chamber orchestra it's not like a lamborghini and doesnt sound like the LSO to me ( unlike what they say on their website)

[JonP]

I watched the 07 namm video about DVZ and despite it sounds like a good chamber orchestra it's not like a lamborghini and doesnt sound like the LSO to me ( unlike what they say on their website)

Proof of the pudding....

The concept sounds quite brilliant to me so I say just relax with what you've got and get ready to be impressed.

[JohnCarter]

Proof of the pudding....

The concept sounds quite brilliant to me so I say just relax with what you've got and get ready to be impressed.

Sure the concept of building the size of the orchestra and being completely free artistically is very good. And i hope we'll have REAL demos soon and seriously these 2 videos are sonically not impressive at all, to me . Perhaps because it's only some strings lines improvised on a keyboard ...

[spectrum]

I really don’t think demos, or the sound, is the important thing here. For one thing they are promising integration with all the major libraries out there (ReMap), and you already know what they sound like. Even if their own instruments sound like crap (and I bet they won’t), that is not the important thing either.
The important thing here is “The Idea”. In this case, “The Idea” is the solving of many of the remaining problems of dealing with multi-articulated sample based instruments, and delivering a streamlined workflow from musical idea to finished composition with a great reduction of the usual effort involved. This promises to be a unified platform package where many long desired features are under one roof: real time articulation changes, distributed processing, real humanizing, divisi sizes (number of instruments playing at once) does not change no matter mow many notes are being played, realistic glide, variable room size with player positioning, real-time pitch manipulation, forced transposition, scoring program integration and more.

Hmmmm....that's not how it looks to me.

I think you might have way too high expectations for what AI will be able to achieve with existing libraries. The way I understand their system, you would NOT gain the following features when integrating existing libraries with their system:

• Real time articulation changes (features/limitations of external library only)
• Real humanizing
• Divisi sizes (number of instruments playing at once) does not change no matter mow many notes are being played
• Realistic glide
• Variable room size with player positioning
• Real-time pitch manipulation
• Scoring program integration (same limitations as using external libraries)

Somebody feel free to correct me if I'm wrong, but I'm pretty sure that those new features listed above can ONLY be achieved with their sounds, because of the way they are sampled and programmed. So indeed, the sound of their samples matters a great deal! :-)

You would, however, most likely have the ability to do the following with external libraries:

• Forced transposition
• Putting external libraries through their SPACE processor

But that's not super different than what you can already do with existing software, reverbs, etc.

So I guess we'll see! :-)

spectrum

[JonFairhurst]

Certainly the copy protected libs can only be integrated so far.

[artinro]

Hmmmm....that's not how it looks to me.

I think you might have way too high expectations for what AI will be able to achieve with existing libraries. The way I understand their system, you would NOT gain the following features when integrating existing libraries with their system:

• Real time articulation changes (features/limitations of external library only)
• Real humanizing
• Divisi sizes (number of instruments playing at once) does not change no matter mow many notes are being played
• Realistic glide
• Variable room size with player positioning
• Real-time pitch manipulation
• Scoring program integration (same limitations as using external libraries)

Somebody feel free to correct me if I'm wrong, but I'm pretty sure that those new features listed above can ONLY be achieved with their sounds, because of the way they are sampled and programmed. So indeed, the sound of their samples matters a great deal! :-)

You would, however, most likely have the ability to do the following with external libraries:

• Forced transposition
• Putting external libraries through their SPACE processor

But that's not super different than what you can already do with existing software, reverbs, etc.

So I guess we'll see! :-)

spectrum
I think Spectrum has hit the nail on the head! Unless "outside" samples are recorded to specifically take advantage of AI's interface, you'll not be able to utilize most features that AI's actual samples do (execpt for space etc...). So, in fact, the quality of their sound is, without question, the most important factor for most of us following this product. DVZ technology is a great idea, but the realism of the sound is much, much more important. As it stands, divisi is already possible to achieve by simply splitting out divisi lines and utilizing smaller ensembles (like VSL chamber etc..) then returning to a larger ensemble during tutti sections.

My guess is that AI is aware of the importance of their "sound" (perhaps they've even read some of these posts) and that this is why actual demos are still not available. They realize that they'll really have to blow people away for this to be a truly viable option for people. I have to say, i've been a bit dissapointed by some of the post-NAMM negative reactions i've read from people who have had a chance to actually see this product in action....but i remain cautiously optimistic that they will take the steps needed to improve on some of shortcomings others have pointed out.

[JonFairhurst]

It's certainly too early to judge AI's products - even for the people who got to hear it at NAMM. (Is there a worse environment for critical listening?)

I read a post recently from one of the VSL guys who looked back at all of the naysayers while they were in development. And look where they are today?

Hopefully AI has enough resources to get the first release "right enough", and then be able to keep developing until their products are killer.

They won't do much to bring the costs down, but if they raise the bar for quality and features, that will all trickle down before long. And we will all benefit.

[dwdonehoo]

From what I have been told by Audio Impressions, the 'modules' from 3rd party libraries will be created by Audio Impressions, not by the 3rd party libraries. I don't know how that will work: I was expecting they would have some kind of SDK for developers (and maybe they will-I don't know). So we will have to see just how deep the integration will be. We really do not know how this will work out until it is released (soon I am told). Certainly skepticism is warranted while hoping for the best no matter who is taking on the workflow 'hairy beast' in software.
Eric's and others concerns are valid: Still, I am very intrigued to see that AI and others are tackling in a unified way the remaining issues of large scale virtual orchestrations.

[tgfoo]

The DVZ engine looks really cool. I definitely like the direction that it's headed in and hopefully others will follow with technology similar to this. As for the samples, I'm hoping the sound top notch (when it's all finished and proper demos are around). However, it wouldn't surprise me if they aren't the best package of samples. From the first video they did and some of the other statments I've heard from Chris Stone it doesn't seem like they designed this with the intention of the samples being used for the final product, but more as a mock up tool during the phases of composition (to later be played by a real orchestra).

[spectrum]

From what I have been told by Audio Impressions, the 'modules' from 3rd party libraries will be created by Audio Impressions, not by the 3rd party libraries. I don't know how that will work: I was expecting they would have some kind of SDK for developers (and maybe they will-I don't know). So we will have to see just how deep the integration will be. We really do not know how this will work out until it is released (soon I am told). Certainly skepticism is warranted while hoping for the best no matter who is taking on the workflow 'hairy beast' in software.

It's pretty simple thing though....unless a sample library is recorded with a "per stand" approach like AI is, none of those more interesting features could work with their system. Since there aren't really current ensemble libraries that are recorded that way, I wouldn't be too optimistic about those features working with any of the current sample libraries on the market.

WHY?

If a single audio sample is recorded with 16 players playing simultaneously in a room, you cannot make that into a divisi 'sub-parts'. AI's approach doesn't work that way either. They are simply recording one or two musicians at a time and then using a ton of computers to fire all the necessary voices to play them simultaneously to produce an ensemble.

When was the last time that overdubbing individual musicians to multitrack sounded like an orchestra performing in a room? :-)

We'll see if they pull it off...they would be the first to do it. :-)

[Garritan]

It's pretty simple thing though....unless a sample library is recorded with a "per stand" approach like AI is, none of those more interesting features could work with their system. Since there aren't really current ensemble libraries that are recorded that way, I wouldn't be too optimistic about those features working with any of the current sample libraries on the market.
AI has some very fascinating techology in many areas and they are to be commended for their innovative thinking. You may not be aware that some of the AI features may work with GPO because both share a similar approach to orchestral sampling.

Personal Orchestra was uniquely developed as an orchestral library comprised of individual instruments, rather than a traditional ensemble library. We took a different approach in which you can build your own ensembles the way you want from individual instruments. So much orchestral literature that has been realized with GPO, not possible to with traditional section libraries, is a testament to this.

Being able to do divisi has been a feature that we have emphasized since GPO was introduced in 2003. This linear individual-instrument approach to orchestral development makes more sense and it is surprising other developers have not adopted this approach. We are glad Audio Impressions has followed our lead in this regard and hope other libraries follow this approach in the future. I believe the traditional sectional approaches may eventually be be rendered obsolete as music technology develops (IMHO).

It seems that many of the AI features could work well with Personal Orchestra (and the upcoming GPO Advanced). I have been in touch with Chris Stone and the folks at Audio Impressions as we share similar philosophies and believe that we can work together.

Gary Garritan

[alprazolam]

.....they sound "keyboard like"......

[spectrum]

AI has some very fascinating techology in many areas and they are to be commended for their innovative thinking. You may not be aware that some of the AI features may work with GPO because both share a similar approach to orchestral sampling.

Personal Orchestra was uniquely developed as an orchestral library comprised of individual instruments, rather than a traditional ensemble library. We took a different approach in which you can build your own ensembles the way you want from individual instruments. So much orchestral literature that has been realized with GPO, not possible to with traditional section libraries, is a testament to this.

Being able to do divisi has been a feature that we have emphasized since GPO was introduced in 2003. This linear individual-instrument approach to orchestral development makes more sense and it is surprising other developers have not adopted this approach. We are glad Audio Impressions has followed our lead in this regard and hope other libraries follow this approach in the future. I believe the traditional sectional approaches may eventually be be rendered obsolete as music technology develops (IMHO).

It seems that many of the AI features could work well with Personal Orchestra (and the upcoming GPO Advanced). I have been in touch with Chris Stone and the folks at Audio Impressions as we shares similar philosophies and believe that we can work together.

Gary Garritan

OK...GPO excepted. Thanks Gary. :-)

Cheers,

spectrum

[Mike Greene]

I think it would be nice to see someone who heard this at NAMM post their impressions! Have there been any threads like that I could check out? :confused:

[Marcussen]

AI has some very fascinating techology in many areas and they are to be commended for their innovative thinking. You may not be aware that some of the AI features may work with GPO because both share a similar approach to orchestral sampling.

Personal Orchestra was uniquely developed as an orchestral library comprised of individual instruments, rather than a traditional ensemble library. We took a different approach in which you can build your own ensembles the way you want from individual instruments. So much orchestral literature that has been realized with GPO, not possible to with traditional section libraries, is a testament to this.

Being able to do divisi has been a feature that we have emphasized since GPO was introduced in 2003. This linear individual-instrument approach to orchestral development makes more sense and it is surprising other developers have not adopted this approach. We are glad Audio Impressions has followed our lead in this regard and hope other libraries follow this approach in the future. I believe the traditional sectional approaches may eventually be be rendered obsolete as music technology develops (IMHO).

It seems that many of the AI features could work well with Personal Orchestra (and the upcoming GPO Advanced). I have been in touch with Chris Stone and the folks at Audio Impressions as we shares similar philosophies and believe that we can work together.

Gary Garritan

However the only two people on the planet owning both AI and GPO will be you and Chris Stone. Honestly if AI can live up to the hype and its price why add GPO?

AI will really have to prove themselves - their setup costs and library costs will make it a very high-end library. The question is then how it will compare to VSL. Someone commented on how AI is meant for being a mockup tool - If true thats a BLOODY expenssive mockup tool - and then, why not just use GPO, Silver or some of KH's stuff? AI needs to be more than that - it needs to be mindnumbingly good.

[Garritan]

Honestly if AI can live up to the hype and its price why add GPO? GPO could benefit from the auto-divisi process. In GPO, divisi is done manually rather than automatically. The AI approach could be a real timesaver. Also, this could also be very beneficial for live performance.
AI will really have to prove themselves - their setup costs and library costs will make it a very high-end library. The question is then how it will compare to VSL. Time will tell. In hindsight, the question may be how will all other libraries measure up to the features in AI. We just don't know yet. Sample libraries are not absolutes in and of themselves and professionals consider them tools in the palette. Some of the composers I spoke with intend to purchase this library. Many of the prospective buyers for this library will have all the libraries and consider it another worthwhile tool.

Gary Garritan

[Garritan]

OK...GPO excepted. Thanks Gary. :-)

Cheers,

spectrumEric,
Of course there's always room for more. ;) The world could use an Eric Persing Mahavishnu Electric-Light Trans-Siberian Orchestra. :D

Best,

Gary

[Marcussen]

GPO could benefit from the auto-divisi process. In GPO, divisi is done manually rather than automatically. The AI approach could be a real timesaver. Also, this could also be very beneficial for live performance.
Gary Garritan

Yup. But I think developing something like that yourself, or licencing their technology would be a more reasonable approach, than hoping GPO users have a 6 PC AI setip :D But I'm sure you have your eye on the ball :cool:

I was just pointing out how I doubt any real world users of AI would put GPO into its engine. Which brings us back to Erics point - it's limited what wonders AI will work for excisting libraries.

[Garritan]

...someone could probably do the divisi trick using GPO and a kontakt2 script. I'm not a programmer, but having seen what scripting can do in libraries like the Strad or even the Akoustic Guitar that comes with Kontakt 2 I would think this would be pretty easy to do. Specially combining all the solo GPO strings with that script that created multiple 'parts' by internally pitch shifting and transposing. That would give you quite a few solo string voices from GPO itself.
Tom,

It would be pretty easy to do. There are different ways to do divisi and the AI's approach seems like a very good approach. There are plugins/programs that have been around for some time that can assign midi channels from chords (or multiple notes) to create a divisi. People have been doing this with Bidule and splitting out chords can easily be done with Kontakt 2 scripting.

But the AI system seems to have a specialized and sophisticated proprietary processes that does specific functions. AFAIK, the AI auto-divisi is a specific algorithmic process that puts specific intelligence into when, where and which particular instruments in their library are used. This is where it shines. It was interesting seeing it demonstrated at NAMM.

Some divisi functions could be done with scripting, but some of the specialized higher-order algorithmic processes probably could not and requires a great deal of processing power. We'll have to wait for AI to provide more details.

Gary Garritan