15 mic inputs driving realtime visuals?

[quickSnail]

Hi,

I'm looking for any comments or suggestion of how I can create a visual performance piece that responds to 15 mic inputs. I'm thinking Max/MSP seems to be the way to go for this, but I'm not sure if max would be able to handle this.

Basically I need a better understanding of how work much processing power it takes to analyze 15 signals simultaneously in realtime and what software/hardware solutions I might need.

thanks

quick:)

[phunkyguy]

i'd bring them all via stereo mix. i don't know your situation, but it seems easier to me.

another options is to use a sound-to-midi dsp plugin and have it spit out midi

[quickSnail]

I need each of the inputs to be isolated, so putting them all on one stereo feed won't work for me. I was thinking about the analog audio to MIDI solution. how much processing power do i need to analyze 15 signals in realtime? And secondly how would i get 15inputs into 1 pc?

[phunkyguy]

you're probably going to need an extremely pricey audio card of some kind. or a mixer.

[elbows]

For a start can we explore in a bit more detail what sort of analysis you want to perform on each mic input?

Depending on what exactly you want, it may or may not be possible.

I would think that even relatively simplistic analysis will be more than 1 pc can handle, and you will need an expensive soundcard or two to get the seperate channels in.

Anyway its something Im interested in generally speaking, can you give more detail about how the mics will be driving the visuals?

One starting point is software like this which can do realtime audio->midi stuff:

http://www.digital-ear.com/digital-ear/index.html

[quickSnail]

We'll I'm just exploring my options really - trying to get an idea of what is feasible and what isn't - whats super expensive , technically difficult etc.

But If its going to take 1pc just to process one audio feed then maybe analog to midi is not the right way.

my bare requirements for triggering of the visuals could be really simple. I would just need to analyze the amplitude of each audio signal and convert it to midi or any other kind of variable from 1 - 100. midi would be good cause then i could use apps like max/jitter to power the graphics. i'm thinking I would use abstract shapes that would change in size or color etc depending on the 15 variables.

[elbows]

Well software like I mentioned above can manage a lot more than amplitude monitoring, so theoretically speaking I would imagine a fast computer could handle plenty of realtime amplitude->midi channels at once. Im not sure however if a suitable/optimised software solution exists for that.

Alternatively do any of the max/msp like environments (eg pd/gem, eyesweb, vvvv) support any kind of audio analysis, maybe, Im not sure.

Have you got a specific mic in mind for this project? If it needs phantom power for example, this will change the hardware requirements (I only just recently bought my first mic and learnt a little about this stuff).

I was interested in midi wind instruments a while ago, as an alternative to a possibly-similar idea to yours, but was thwarted by an apparent end to midi wind instrument market within the EU for reasons I didnt discover.

I dont know enough about how mic/audio signals work electrically/electronically, but amplitude detection and conversion to midi via relatively simple hardware box sounds plausible to me, potentially much more robust than software too.

Im not currently doing anything in this area myself, being far too distracted by 100 other things Ive started and never finished, but if it were me Id probably play with a few different audio->midi softwares and a cheap mic plugged into a normal soundcard, and see if the midi data you get from it is stable enough to create visuals that match whats in your mind. Ive aimed for too precise control in the past, and ended up dissapointed when the inevitable noises/glitches of one kind or another cause unwanted disturbances in the visual result, but this problem for me could be turned into a positive advantage with clever choice of how the visual elements are influenced by the mics, and for more abstract stuff.

Part of my current agenda is using different realtime 3D software to generate and/or influence abstract and non abstract images controlled by midi, but in my case the midi will be coming from typical midi controllers and midi sequencing software. Im currently fascinated by how many elements can be influenced by different sources at once before the brain overloads and cant process it all quick enough to notice how the audio and visuals are relatign to eachother. Even with relatively simple changes, like size and colour that you mention, Im finding that less is more.

Anyway sorry for going on about me, shouldnt be hijacking your thread, just hope some of it is vaguely relevant.

[signal55]

you could try visual jockey i find the audio synch mode quite dynamic and quite accurate in its response's to the different audio frequencies!

xxmossy

[many2]

Unless you can convert the music into 15 different midi signals first visualJockey isn't the solution : visualJockey's music analysis is done on one channel only, not 15. With a music to midi system though it would be a very flexible tool to control visuals with 15 (or more) controlling channels.

I am experimenting with similar things at the moment but with direct MIDI control - very powerful.

[signal55]

yeah i know, but because you can synch various aspects of a composition to different audio frequencies, i suppose in theory you could have a filter or eq on each mic before mixing down and sending it to visual jockey though i've only ever tried using a eq on one channel before!

xxmossy

[sleepytom]

max/msp is probably the most advanced "off the shelf" solution for this - with a multiple input sound card you should be able to manage 15 mic inputs.

[levon]

i know there are hardware audio-midi, but never used one before so i dont know.

one thing about trigering visuals, you would only need a low sampling rate much less then the 44khz that most sound cards could do, there is some way to carry multiple signals of a smaller frequency over 44khz, though i would have no idea how to do it and you would have to make the hardware and drivers/software you're self.

[Feldspar]

I dont know how much CPU would be used for 15 channels, but I'd think the hard part would be getting a sound card with 15 inputs. Using an EQ or bandpass filter you could probably combine some signals that dont ovelap much on frequency without losing any accuracy. Whatever software you use would then have to respond to isolated frequency bands in the combined signal.

[tdeagan]

Just for a different perspective...

As a hardware hacker I would approach this problem differently.

So far the requirements have only discussed needing basic amplitude info to trigger from the mics. I'd investigate one of the following approaches;

1) 15 channels of ADC (analog to digital conversion) feeding into a fast Basic Stamp (two MAX 186 8-ch ADCs into a BS2, something along these lines;
http://users.minet.ca/tcumming/public_html_renamed/Electronics/circuits/max186.html ) that then comes in through serial or MIDI ( http://www.jordanautomations.com/basicstampmidi.html ) Code in the Stamp could do FFTs or any number of interesting things other than just amplitude.

2) 15 simple level triggers (could be something like http://www.paia.com/drumsens.htm or even cheaper with research) that feed into a parallel port input read by VB code or somesuch.

3) some crazy deal that builds a little battery powered box for each mic that then outputs IR signals all collected back at the PC by a Basic Stamp monitoring IR and outputting a MIDI signal

In any event, once you're into doing something with 15 mics, you've crossed the rubicon either into complexity or cash.

I'd end up with the first solution, it's a lot easier than it looks and would be a pretty cool piece of gear to have.

[alangeering]

Totally agree, I was about to sugges a similar plan. Really it comes down to:

If only measuring Amplitude:
Resolution: {specified} 1-100, simple
Sample Rate: {guessing} 30 Hz
Solution: Microcontroller based design

If measuring Amplitude at various frequencies:
Resolution: {specified} 1-100, simple
Sample Rate: {guessing} 22 to 44 KHz
Solution: High end sound card or DSP based solution.

The amplitude only solution could be done with some PICs or BASIC stamp (as has already been suggested). As the microcontroller can switch inputs at a rate higher than 30 Hz then it would be feasable to do this with 2 or 3 microcontrollers (i.e. 8 lines of A/D conversion per microcontroller, 2 of them, plus a possible other to handle comms)

I'm way too busy to design a circuit right now but you can find helpful hits through google, it's been done before.

[CA]

I dont know what your budget is but this might be a hardward fix

http://www.aardvarkaudio.com/products/q10/

[hamageddon]

since the problem here is getting 16 Audioinputs into a 'puter
u mite consider throwing ur money towards a modular hw solution like the doepfer A100 (http://www.doepfer.de/home_e.htm)

i suggest a combination of a bunch of A119s (http://www.doepfer.de/a119.htm) and a
A192 (http://www.doepfer.de/a192.htm) here

shouldn't be as expensive as anA330 (http://www.airbus.com/product/a330_a340_backgrounder.asp) :D

[quickSnail]

hey all,

thanks for all the info, Its funny I thought this post was dead, I just got a notice of all these replies - now a week later.

anyways, It looks like I'll be going with a custom designed audio to midi box. A friend of mine will design and build them - once there built we will plug each device into a MIDI USB interface and then use either MAX/MSP, TOUCH or PD. I'd like to have the option of realtime 3D for the visuals so I'm thinking Touch might be the best although priciest solution as far as visual quality and ease of use goes for the motion graphics authoring.

All this was for a grant proposal for doing visuals to an 15+ orchestra. I've got my fingers crossed.

If this wasn't a grant proposal, and if I had to do it without the custom midi box and on the cheap, I think I'd use the Teleos system and MAX/MSP or FLash. The Teleos analogue in module
can hook up to a computer via USB and allows 12 analogue inputs. I would probably have to modify the mic signal to a lower voltage but that deosn't seem that difficult.



Actually I've never heard anyone mention the Teleos system anywhere on these forums. so FYI heres the module i was talking about http://www.makingthings.com/products/documentation/teleo_Analog_in_user_guide/index.html

[levon]

be sure to keep us updated on your progress on this, and it would be great if you could post the schematics and that once it is done

good luck

[acousmod]

Hi,

Perhaps that this will help you : I've made a VST plugin that can analyze and convert two audio inputs to two MIDI cc.
As a VST plug you can use 8 of them in parallel in a host like Synthedit or Audiomulch (freewares)

You can get it at :
http://acousmodules.free.fr/index.htm
see at MIDI Controllers / MIDIPlugs

Concerning the audio input, obviously you will need a sound card with 16 inputs, which costs a lot, and often with line inputs only..

See at MOTU or RME

cheers

[dolabriform]

Hi all

Many musical instrument manufacturers have made audio to MIDI converters, AKAI, Roland, and Alesis have all made boxes that convert a voltage to a MIDI note or controller.

With a pre-amp on the front end, you should be able to get enough voltage from a microphone to power the trigger.

Many of these boxes are now available second hand for little money, and most have 8 inputs.

I have a Roland TD5 that I use regularly for triggering, and the Alesis DM5 will also do the trick (this has 12 trig inputs).

Both where originally designed for triggering drum sounds, but I use them for all sorts of things!

I hope that helps

D :-)