What's new

Sample Modeling has been busy....

I like music

Senior Member
I need striiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiings!!

Ok, sorry. :confused:
Anyone know how long sample libraries stay in beta? If we assume that at the latest, SM strings went into beta in Feb some time, how far are we from release? I know that's a bloody hard one to guess, but why not conjecture? It is fun. Any takers?
 

DANIELE

Active Member
Anyone know how long sample libraries stay in beta? If we assume that at the latest, SM strings went into beta in Feb some time, how far are we from release? I know that's a bloody hard one to guess, but why not conjecture? It is fun. Any takers?
I was just asking myself when their library will be out. They wrote "New demos will be out soon" but today (June) we still haven't heard any other things about it.
I don't want to bother anyone but I'd really like to know something about them since I'm missing playable strings so much in my template.
 
I was just asking myself when their library will be out. They wrote "New demos will be out soon" but today (June) we still haven't heard any other things about it.
I don't want to bother anyone but I'd really like to know something about them since I'm missing playable strings so much in my template.
Dear friends,

We are very happy to see the enthusiasm for our upcoming string library.


All the instruments, solo & ensemble, are in a final stage of beta testing and fine-tuning. Please consider that it is a very complex project, encompassing solo violin, viola, cello, double bass, and expressive ensembles based on our proprietary technology.
As we break new ground during the development process, implementing changes in all the instruments turned to be a very challenging and time consuming process. We believe that our results will justify the additional time requested for development. The instruments will be shortly in the hands of Native Instruments for encoding. It wouldn't take too long. Thanks for your patience and encouraging expectations.

Stay tuned.

The Samplemodeling Team.
 
Last edited:

I like music

Senior Member
Dear friends,

We are very happy to see the enthusiasm for our upcoming string library.


All the instruments, solo & ensemble, are in a final stage of beta testing and fine-tuning. Please consider that it is a very complex project, encompassing solo violin, viola, cello, double bass, and expressive ensembles based on our proprietary technology.
As we break new ground during the development process, implementing changes in all the instruments turned to be a very challenging and time consuming process. We believe that our results will justify the additional time requested for development. The instruments will be shortly in the hands of Native Instruments for encoding. It wouldn't take too long. Thanks for your patience and encouraging expectations.

Stay tuned.

The Samplemodeling Team.
Yes!
 

Sam Reed

stranger in a strange land
Dear friends,

We are very happy to see the enthusiasm for our upcoming string library.


[...] We believe that our results will justify the additional time requested for development. [...]
Dear Giorgio,

Congratulations on the most recent undertaking! As a longtime user of your Brass instruments, I have come to expect great things from you, Peter and the rest of your team.

When might we expect an announcement with a bit more details? Specifically:

1) a rough price estimate so we know how much to set aside
2) the size(s) of ensembles which will be possible. The only demo so far is of a small ensemble; will this release be best for chamber-sized ensembles or will it encompass "solo to string quartet to 16-14-12-10-8 and everything in between" (here's hoping for the latter! ;) )
3) will the initial release include similar features as were added to the brass instruments in the 3.0 versions? (e.g., ERs and IRs specifically tuned to the instruments, ensemble IRs, etc.)

Kind regards,
Sam
 
Dear Sam,

thank you for appreciation and your interest in our products. :) Here are some further details:

will the initial release include similar features as were added to the brass instruments in the 3.0 versions? (e.g., ERs and IRs specifically tuned to the instruments, ensemble IRs, etc.)
Yes. These new Samplemodeling Strings (sample-modeled based) will offer all the features of Samplemodeling Brass, and much more.

the size(s) of ensembles which will be possible. The only demo so far is of a small ensemble; will this release be best for chamber-sized ensembles or will it encompass "solo to string quartet to 16-14-12-10-8 and everything in between" (here's hoping for the latter! ;) )
The size(s) of the ensembles must be viewed from a different perspective. This is not a conventional library, so the approach of sampling, for example, 8 vs. 12. vs 16 instruments does not apply here. We started from different concepts and adopted a quite unconventional approach. The user will be able to adjust, in a continuous manner, the perceived ensemble size from "small" to "large".

a rough price estimate so we know how much to set aside
The product will be on sale only as a complete set (solo & ensemble strings, including first and second violins, violas, cellos, and double basses). The price will be 399 EUR plus an additional time-limited discount of 10 % for all our Samplemodeling customers, as usual.

Peter & Giorgio
 

Nicola74

Member
Can't wait to see this library in action!!
Any prevision about the release date? Just to understand what means "coming soon" ;)
 

maestro2be

Active Member
I can't remember the name of the product but many years ago a company tried to do a modeled string library that you could click on string icons one at a time to increase it's size to make your own. Sounds like this might be a similar approach from the hints. Be interesting to see some walkthrough when it's ready.
 

I like music

Senior Member
I can't remember the name of the product but many years ago a company tried to do a modeled string library that you could click on string icons one at a time to increase it's size to make your own. Sounds like this might be a similar approach from the hints. Be interesting to see some walkthrough when it's ready.
Was that DVZ strings?
 

Sam Reed

stranger in a strange land
Thank you, Peter, for the excellent information! It all sounds so promising, and based on your track record, I don't foresee any reason not to become an early adopter.

The size(s) of the ensembles must be viewed from a different perspective [...] The user will be able to adjust, in a continuous manner, the perceived ensemble size [...]
I don't understand the full ramifications of viewing the size of the ensembles "from a different perspective" (I always think from a real world perspective, having little understanding the details of how libraries are created), but I'm sure you folks will explain this quite well in the manual (which was certainly my experience with Samplemodeling Brass' manuals, even for concepts which were new to me at the time). Looking forward to more information about this when the time comes.

"Continuous" control of perceived ensemble size sounds like a dream when writing divisi!
 

filcho89

New Member
What I'm hoping for in this instrument is simply a solution for a few more problems. I'm 100% on board with the approach (modeling samples instead of 3d instruments).

On one hand, the demo seems to show that realism is not going to happen with the solo instruments. The demo features the solo violin, which was a bold move, and honestly, it caused me to be discouraged at first. Lots of strange artifacts in the transitions between notes (almost resembles digital distortion), and the different aspects of the sound all seem way off. I don't hear realistic variations in bow speed, bow pressure, bow stick (I'm referring to the start of a stroke), current string, or any aspects of the fingering, note connectedness, etc. In addition to that, imperfection is important wherever realism is the goal, so that would need to be programmed in somehow, especially concerning intonation. The physical modeling approach addressed practically all of these aspects far better so far. I've attached an example of a programmed solo line using physical modeling. The tone of the instrument is a bit off, as are some of the note connections and the sound of the rosin. But in my opinion, it gets us far closer than the Vivaldi demo.

How will Sample Modeling address these things? I'm eager to see and hear.

The ensembles are a different ball game. They may very well sound better than the solos. It may be possible to create a real-sounding ensemble out of fake sounding members. I bet some types of lines will be possible that have never been possible before. That's exciting to me. I think modeling samples is the way to make seamlessly playable string sections. Physically modeling a string section is still fairly utopian, because it may be 30 years before a $1000 computer can process a 50-100 physical models all at once. This is why Audio Modeling will probably not offer us feasible solutions for string section writing anytime at all soon.
 

Attachments

alanb

Senior Member
That video takes music into the uncanny valley. Google it. I personally found it disturbing. The track by itself is impressive, but the concept of overwriting a real performance is somehow creepy. IMHO.
I don't find it creepy but I understand that it fools the ear better when you see someone actually playing the instruments.

It's never creepy when the 'overwritten' performance is a StSanders Production!!!


[or even a halfway-decent imitation, like I believe the last one is . . . in any event, it's got yr emotional 'cello right there...]






 

I like music

Senior Member
What I'm hoping for in this instrument is simply a solution for a few more problems. I'm 100% on board with the approach (modeling samples instead of 3d instruments).

On one hand, the demo seems to show that realism is not going to happen with the solo instruments. The demo features the solo violin, which was a bold move, and honestly, it caused me to be discouraged at first. Lots of strange artifacts in the transitions between notes (almost resembles digital distortion), and the different aspects of the sound all seem way off. I don't hear realistic variations in bow speed, bow pressure, bow stick (I'm referring to the start of a stroke), current string, or any aspects of the fingering, note connectedness, etc. In addition to that, imperfection is important wherever realism is the goal, so that would need to be programmed in somehow, especially concerning intonation. The physical modeling approach addressed practically all of these aspects far better so far. I've attached an example of a programmed solo line using physical modeling. The tone of the instrument is a bit off, as are some of the note connections and the sound of the rosin. But in my opinion, it gets us far closer than the Vivaldi demo.

How will Sample Modeling address these things? I'm eager to see and hear.

The ensembles are a different ball game. They may very well sound better than the solos. It may be possible to create a real-sounding ensemble out of fake sounding members. I bet some types of lines will be possible that have never been possible before. That's exciting to me. I think modeling samples is the way to make seamlessly playable string sections. Physically modeling a string section is still fairly utopian, because it may be 30 years before a $1000 computer can process a 50-100 physical models all at once. This is why Audio Modeling will probably not offer us feasible solutions for string section writing anytime at all soon.
Interesting. By the way, did you use SWAM violin there, or something you've developed/are developing?
 

filcho89

New Member
Interesting. By the way, did you use SWAM violin there, or something you've developed/are developing?
It is in fact SWAM. It's not performed though; I edited the values individually, mostly on separate CC lanes:

-Bow speed
-Bow pressure
-Bow position
-Bow change
-Release
-Attack
-Fingering (including choice of position/string)
-Note connectedness
-Vibrato speed
-Vibrato intensity
-Intonation imperfections (usually starting a note slightly out of tune and adjusting, increasing imperfection according to register, playing speed, interval length, and connection type)
-I also added a little EQ to bring out the rosin (but I may have missed the mark a bit)

I'm sure Sample Modeling is tired of being associated to Audio Modeling's products, especially since they're pursuing a different approach so deliberately. In terms of the future of the technology, modeling samples is more interesting to me than virtual objects, but for the time being, the virtual objects provide us with a neat way of studying player behavior and aspects of the sound profile. I'm especially interested in the player behavior part.

While I'm a little concerned that the new instrument might not have gotten the sound profile quite right, I'm optimistic. It's an instrument that potentially could have a more linear progress over the course of the years following release, unlike sample library-based instruments that are forever married to, and limited by, the initial recordings, as well as the vision with which they were recorded.

If the SM developers read this, here's my two cents:
Unless every aspect of the sound becomes dynamic, the new approach loses its meaning. Not only that, but each aspect of the sound must be dynamic in a way that reflects real player behavior. I've always been disappointed that both Sample Modeling and Audio Modeling focus on making an instrument "playable". The goal should not be to give the consumer an instrument that is 100% controllable. We, the consumers, are not smart enough to understand the nuances of realistic violin performances. We need SM to automate characteristic of bow pressure, for instance, as well as the minor nuances of intonation, and much much more! If you don't do this, I suspect Note Performer will, in a matter of a few years, become a much more viable way of creating realistic string performances. Artificial intelligence is absolutely the path upwards.
 

DANIELE

Active Member
It is in fact SWAM. It's not performed though; I edited the values individually, mostly on separate CC lanes:

-Bow speed
-Bow pressure
-Bow position
-Bow change
-Release
-Attack
-Fingering (including choice of position/string)
-Note connectedness
-Vibrato speed
-Vibrato intensity
-Intonation imperfections (usually starting a note slightly out of tune and adjusting, increasing imperfection according to register, playing speed, interval length, and connection type)
-I also added a little EQ to bring out the rosin (but I may have missed the mark a bit)

I'm sure Sample Modeling is tired of being associated to Audio Modeling's products, especially since they're pursuing a different approach so deliberately. In terms of the future of the technology, modeling samples is more interesting to me than virtual objects, but for the time being, the virtual objects provide us with a neat way of studying player behavior and aspects of the sound profile. I'm especially interested in the player behavior part.

While I'm a little concerned that the new instrument might not have gotten the sound profile quite right, I'm optimistic. It's an instrument that potentially could have a more linear progress over the course of the years following release, unlike sample library-based instruments that are forever married to, and limited by, the initial recordings, as well as the vision with which they were recorded.

If the SM developers read this, here's my two cents:
Unless every aspect of the sound becomes dynamic, the new approach loses its meaning. Not only that, but each aspect of the sound must be dynamic in a way that reflects real player behavior. I've always been disappointed that both Sample Modeling and Audio Modeling focus on making an instrument "playable". The goal should not be to give the consumer an instrument that is 100% controllable. We, the consumers, are not smart enough to understand the nuances of realistic violin performances. We need SM to automate characteristic of bow pressure, for instance, as well as the minor nuances of intonation, and much much more! If you don't do this, I suspect Note Performer will, in a matter of a few years, become a much more viable way of creating realistic string performances. Artificial intelligence is absolutely the path upwards.
Well I would like to have the possibility of having some aspects automated but also the option to have them controllable if I want.
 

Fa

Member
What I'm hoping for in this instrument is simply a solution for a few more problems. I'm 100% on board with the approach (modeling samples instead of 3d instruments).

On one hand, the demo seems to show that realism is not going to happen with the solo instruments. The demo features the solo violin, which was a bold move, ...l. The physical modeling approach addressed practically all of these aspects far better so far. I've attached an example of a programmed solo line using physical modeling. The tone of the instrument is a bit off, as are some of the note connections and the sound of the rosin. But in my opinion, it gets us far closer than the Vivaldi demo.
Dear Filcho, if we talk about research and technology approach, I agree with everything you wrote. I’ve also good news, because actually the new SM product is doing a lot of what you are looking for and descrbing, and something more.

But if we talk about the demo, sorry but I don’t understand your point. Several of the sequence nuances were actually my deliberate choices, and have nothing to do with the instrument itself. I accept your criticism, and I apologize if I didn’t render the sequence in the way you would love it.

On the other side I think your sentence is a bit too strong and out of the context: I would invite you to tell me if the sequence and the instrument sound were so inferior to other existing tools, and I would love to listen to the same sequence made in a more realistic way and better sounding. None of my Vi in my large and actual collection can do anything similar, but maybe it’s my fault and anybody is invited to show me he can do it better with another library.

About SWAM as you say it’s comparing apple with pears: I love SWAM, I am a great friend of SWAM producers, and I did a lot of work and demos with it, including swam-s, then I am not a detractor at all, but as you say, it has strength and weakness. I don’t think your sequence is a good benchmark, because it’s just random sounds and noise, and not a fair attempt to make musical good sounding example of my Vivaldi simulation. Your example shows the wonderful power of physical modeling in replicating some of the instrument behaviours, and also the discouraging weakness of the sound and programming complexity. And you agree on that, then I join you in the hope of future development, and I think you will be pleased already by several of the features of the new SM strings VI, and future improvements can just be even more exciting.
 

chapbot

omnivore
It looks like several of the sample modeling staff are on these forums - I think your new ensemble strings technology is really incredible and I am super excited to hear more demos! Do you think you could create a few more in different styles to whet our appetites :grin:
 

DANIELE

Active Member
It looks like several of the sample modeling staff are on these forums - I think your new ensemble strings technology is really incredible and I am super excited to hear more demos! Do you think you could create a few more in different styles to whet our appetites :grin:
I think we could listen to more demos once the library is out. I hope we don't have to wait so much to see it released.
 
Top Bottom