Physical Modeling

  On 8/9/2015 at 3:21 AM, John Ehrlichman said:

Doesn't Kaivo have a demo version?

Yeah it does, and I probably should've tried the demo before asking! It seemed good enough and had I not got the AAS stuff would've been tempted but I don't think it offers me anything more than what I can get with the products I own. The modular routing is quite good fun but I think I've been able to get more useful results from Chromaphone and/or String Studio and/or Tassman

  On 8/5/2015 at 6:02 AM, Mesh Gear Fox said:

why did phys mod seem to be the next big thing in the 90's before dying on its arse, doomed to be explored by smaller companies making vsts and whatnot? not enough public interest?

basically if you want it to sound like a real instrument it takes similar skill level or time automating to articulate like a real instrument. you can imagine how well that went over with people that want to play jump. lol

 

yamaha vl stuff and oasys pci are fun to break and they do so in some enjoyable ways, but kinda the point of them putting tons of time and money into it is that they added all this stuff to make it harder to do that...

 

you can tell that a lot of their effects algorithms in other machines are related because you can get there feeding back and pushing to upper and lower limits. the ability to morph is great, just like with additive, but you can that pull a lot of that off with clever macro programing. there are some interesting note interactions you can get too, but then they never really got the voice count high enough.

 

gen does best stuff i've heard. it's interesting to try to model the space it's in at same time as the instrument so then there isn't really a distinction.

here is something i just put together using all VL1 + a little bit of arturia origin bass at the very end

and older one:

Edited August 24, 2015 by John Ehrlichman

sounds cool, seems like a synthesis technique well worth further exploration

all things considered, I think there is a distinct level of crappiness in the current state of software physical modeling synths.

 

I gave Tassman a good listen, it seems to be the 'best' but it is seriously dated and flat sounding.

 

The VL1 sounds the best of all.

 

The majority of physical modeling plugin instruments are win32 only, synthedit at that.

 

Yamaha, just port the VL1 over to software please, we have the technology.

 

Time to study physical modeling myself, dammit. I really don't have time for this...

  On 8/25/2015 at 3:45 PM, skibby said:

Yamaha, just port the VL1 over to software please, we have the technology.

I was thinking we should get some of the retro arcade/MAME guys together and have them do the same thing with VL1, i.e. copying the ROMs from the chips and emulating the hardware.

Its frustrating that there is all this great tech just rotting away. I think this is the moment I will stop liking Yamaha.

 

Its like how the US went to the moon with a tin can and a casio watch 10 times but now we can only go a few miles maximum up in the air, blowing up spaceships 75% of the damn time.

  On 8/25/2015 at 5:56 PM, skibby said:

Its frustrating that there is all this great tech just rotting away. I think this is the moment I will stop liking Yamaha.

 

Its like how the US went to the moon with a tin can and a casio watch 10 times but now we can only go a few miles maximum up in the air, blowing up spaceships 75% of the damn time.

don't forget that Vocaloid was created sometime around 2003. After the initial development Yamaha completely stopped and instead just made their money putting out different voices. Can you imagine how powerful Vocaloid technology could be with 11 years of additional development to the original algorithms?

Its kind of fascinating how many great tools exist for CGI/visual 'reality simulation' in comparison to reality simulation audio tools. We can now almost get fully realistic looking human characters with all CGI but we're still using 2003 speech synthesis technology as the 'best' ? makes no sense unless you start factoring in capitalism.

 

i agree about the Vl1 (obvio) BUT have you used Sculpture, Kaivo, or Xoxos plugin suite yet? Don't write off software phys mod until you try all of those. If you're still not satisfied buy a VL7 or a VL1. Cheapest way to get the most bang for the buck on yamaha physical modeling is to get 3 x PLG100VL cards and put them in something like a Mu128 (youll end up spending the same amount of money as a VL7 which is only monophonic, whereas 3xPLG cards will give you 3 voices)

Edited August 26, 2015 by John Ehrlichman

  On 8/26/2015 at 2:51 AM, John Ehrlichman said:

 

Its kind of fascinating how many great tools exist for CGI/visual 'reality simulation' in comparison to reality simulation audio tools. We can now almost get fully realistic looking human characters with all CGI but we're still using 2003 speech synthesis technology as the 'best' ? makes no sense unless you start factoring in capitalism.

 

 

Yeah. As expensive as big, summer blockbuster CGI extravaganzas are, CGI is still massively cheaper than filming in actual locations, or with actual special effects. So there is a big market for new technology for CGI before you even factor in games and all the other areas computer graphics are used in.

also part of the reason why physical modeling won't catch on is because sound designers for things like video games would be put out of business. If you could make a game engine that 'generates' things like environmental sounds or gunshots all in realtime, you wont have any need for foley artists or recordists. I know its hard to imagine a world that would employ such a technique but things could have evolved that way, instead visual tools are way farther along

Edited August 26, 2015 by John Ehrlichman

I'm digging those tracks, JE.

  On 8/26/2015 at 9:24 AM, Mesh Gear Fox said:

imagine crafting instruments in blender and adding all these tubes, blowpipes, things to pluck, anything that generates or conveys physical waveforms that we would hear as sound.

Striking & plucking wise Chromaphone has that sorted -

 

 

(The resonators can be a string, beam, marimba, plate, open or closed tube)

  On 8/26/2015 at 9:24 AM, Mesh Gear Fox said:

imagine crafting instruments in blender and adding all these tubes, blowpipes, things to pluck, anything that generates or conveys physical waveforms that we would hear as sound.

Theoretically this could be done with finite element analysis (which is what people use for modeling pressure and strain and flow inside molds and what not, all from CAD models). This guy talks about it a bit near the bottom of the page:

http://www.lumanmagnum.net/physics/real_synth.html

  On 8/26/2015 at 2:44 PM, Mesh Gear Fox said:

edit: $100US

Not my most expensive physically modelled VST though - That'd be Pianoteq 5, but oh my is it worth it ! Even an idiot like me can make a

 

  On 8/26/2015 at 2:51 AM, John Ehrlichman said:

 

  On 8/25/2015 at 5:56 PM, skibby said:

Its frustrating that there is all this great tech just rotting away. I think this is the moment I will stop liking Yamaha.

 

Its like how the US went to the moon with a tin can and a casio watch 10 times but now we can only go a few miles maximum up in the air, blowing up spaceships 75% of the damn time.

don't forget that Vocaloid was created sometime around 2003. After the initial development Yamaha completely stopped and instead just made their money putting out different voices. Can you imagine how powerful Vocaloid technology could be with 11 years of additional development to the original algorithms?

Its kind of fascinating how many great tools exist for CGI/visual 'reality simulation' in comparison to reality simulation audio tools. We can now almost get fully realistic looking human characters with all CGI but we're still using 2003 speech synthesis technology as the 'best' ? makes no sense unless you start factoring in capitalism.

 

i agree about the Vl1 (obvio) BUT have you used Sculpture, Kaivo, or Xoxos plugin suite yet? Don't write off software phys mod until you try all of those. If you're still not satisfied buy a VL7 or a VL1. Cheapest way to get the most bang for the buck on yamaha physical modeling is to get 3 x PLG100VL cards and put them in something like a Mu128 (youll end up spending the same amount of money as a VL7 which is only monophonic, whereas 3xPLG cards will give you 3 voices)

 

 

i think Xoxos should just be given a new identity and a billion dollars so he can go set up shop somewhere far away from the conspiracy that constantly besieges him and hire a team of coders and artists to get things up to speed already.

 

don't like kaivo. the interface just doesnt work for me. if i can't figure it out, i won't even bother trying to use it. the sounds are ok, but i don't care for mixing physmod and grains. it is OK for sound designers I guess, but the aesthetics are just antithetical. physical models mixed with uber digital stuttering in one single voice is too much for me.

 

structure... cant use it cause im on a pc and dont use logic either.

Edited August 26, 2015 by skibby

  On 8/26/2015 at 9:49 AM, psn said:

I'm digging those tracks, JE.

thanks, mostly all of it will be coming out as Ambiente 4, 5 and 6

 

a large percentage of it won't be classical style though, about 50% will be in the style of Metal Machine Music. Massive walls of (fake) guitar noise ambient

You need any real guitar on that?!? :p

 

nice, very old school comb filtering style physmod , and sure if you want to try adding some real guitar over my stuff shoot me an email - peterjacksonitis AAAAATTTTTT geeeemale dot com

  On 8/26/2015 at 4:28 PM, mcbpete said:

 

  On 8/26/2015 at 2:44 PM, Mesh Gear Fox said:

edit: $100US

Not my most expensive physically modelled VST though - That'd be Pianoteq 5, but oh my is it worth it ! Even an idiot like me can make a <30mb piano sound pretty realistic with a few un-quantisation tweaks and some tape emulation:

 

yes you can make a piano sample library sound excellent, as you have :) BUT pianoteq really cannot be replaced with a sampler. There is a fader on the program to 'age' the piano or even turn a regular piano sound into something that almost sounds like a prepared piano with vastly different timbres for every key. Reid/Wisp has done some amazing stuff with pianoteq, not sure if he posted any on soundcloud

Haha, that *is* pianoteq with the aged slider up, not a sampler! (the 30mb I was talking about was the size of the installer )

oh lol, laptop speakers are a hell of a drug

I haven't used physical modelling synthesis in a while, but I did use this vst called Kaivo for this track:

https://soundcloud.com/evoava/chromatic-hush-very-wip

A ton of glitches past 1:55 were ran through it, recorded, spliced, and sewn together. Basically a bajillion 'one shots' thrown together. Takes a while to sequence though.

Edited September 17, 2015 by Brisbot

really nice track. so you're using Kaivo as an effect kinda (running samples through it?)

yep, to be honest I probably shouldn't have even mentioned it because looking back I used it to fatten the samples up a bit (they were super quiet).

Again after 1:50 (its basically the sharp formant sound), it was more important than kaivo that I had a frequency shifter on the samples too, which is what's giving it it's fluid feel, as if I had used a synthesizer for it or something. Without the shifter it would have sounded like a ton of one shots (some repeated a few times in a row, some changing rapidly) and would not have sounded musical.

I basically put together 'large' (few hundred ms) granules that I made individually in a controlled way using my ear, using the freq shifter to make it feel at least somewhat seamless between granules. I wish I could explain it better but it would be way easier to just show you. I've been thinking about making a few tutorial videos on how to make these kinds of sounds. I would like to see what other can do with it.

Edited September 18, 2015 by Brisbot

RE: using 3d models to generate sound, pretty sure it's been possible for a while, they did some of that here with interesting results:

http://www.ness-music.eu/overview/physical-modeling-synthesis

 

Good site to explore, although I wish they would share some source code. Shows some interesting possibilities