Hi All,

I've just released a free app for iOS called Samvada that uses Pure Data, through libpd and pd-for-ios. It's meant to simulate sympathetic strings, like on a sitar, for practicing and performing Indian music. It's also fun for other uses, and the strings make a nice resonance turning background sounds into tones. You can watch the demo and download it here:

Samvada for iOS

It's basically a bank of Karplus-Strong resonators. The UI handles the tuning aspects, which are aimed at Indian musicians using a drone and the Sa Re Ga Ma system of note naming/notation.

I'd welcome any feedback or questions about the process of making an app like this, etc.

Enjoy!

That's an interesting idea acreil. I don't know so much about pitch glides in cymbals, gongs etc, but it occurs to me that something like that might be an interesting way to model another kind of Indian jawari - the jawari on a tabla head. As I understand it, there's a thread that goes round the part of the rim of the tabla that acts in a similar way to the jawari on a sitar or tambura. I don't know if much work has gone into modelling tabla, but it would certainly be an interesting project.

Hey.

• different length delay modulating the string length. Yeah, I guess what you'll get there is like FM sidebands, except with both carrier and modulator resonating according to the input.

• nonlinearity in the return. Yes, I used that idea in Samvada - a cubic soft clip, I think.

• "acousticness" - absolutely. "Physical modelling" sounds impressive but it's really very simple elements. It's amazing how organic the results can be.

Hi acreil. Since I had needed to keep the patch as fast as possible to run 13 resonators on fairly low-performance devices, I cut as many corners as possible. So yes, I used [s~] and [r~] objects to feed the one [vd~] back to its [delwrite~]. No it's not ideal, and you have to keep the "jawari" amount pretty low to avoid some really nasty feedback, but that's the best I could get away with under the performance limitations. I know others have used a model as in your suggestion - two separate delay lines - and this seems a good idea if you have the CPU available.

I don't know if you fixed this hodok. In order to get pitches high enough, i had to use a block~ 16 just around the core K-S synthesis unit.

Hi hodok,

I'm glad you like Samvada - thanks.

Unfortunately it's a bit hard to send you the core string synthesis patch as it contains a custom external (2nd order allpass filter) and currently relies on messages from Objective-C for much of its functionality. However, I can describe what it does. There are two tricks used that add a bit of jawari tone to the basic Karplus-Strong waveguide:

1. Feeding a (small!) amount of the output back into the delay length of the waveguide. This simulates the sitar's flat bridge, slightly changing the length of the string with each vibration. Watch out though, more than even a tiny amount will create the most horrible digital feedback sound ever.

2. String inharmonicity. Adding an allpass filter in the Karplus-Strong loop (I used a second order one) will do this. For some reason the combination of this with point 1 above results in something like the jawari sound (though it's not perfect).

You might want to check http://www.talaash.at/sitar/ which has a great sitar model. I had to cut a few corners in my version: cramming 13 oscillators into an iOS app is pushing things in terms of performance.

Let me know (here or on email) if you have any other questions.

Hi Colorplates,

I'm glad you enjoy using Samvada! I also have a dilruba, though I don't play it much. My main Indian instrument is the bansuri - I made Samvada to help me get to grips with Indian ragas on that.

You are not the first person to ask for Audiobus support - I have a copy of the SDK and am looking into it.

Re: "just intonation", etc. You know you can make your own presets, right? Just tap the 'action' button and then 'Save To Preset ...': you'll be prompted for a name, and if it's a new name it'll make a new preset. I should probably document that somewhere. In the new version you can see how many cents from Sa you are for each string tuner, so if you have a list of cent values for your desired tuning you can just dial them in.

If you come up with useful presets, please let me know - it would be good to be able to share them with other users.

When you say, "the wave that is used to generate the tones" - do you mean the drone, the pluck or the sympathetic resonance? The strings are physically modelled, not samples - but the excitation stimuli (pluck, drone) are samples. Yes it's possible to change them, but I'd rather keep the interface as simple and powerful as possible.

I hadn't thought of making the string plucks able to receive MIDI - I'll look into how easy that might be to add.

Many thanks

Hi all, I'm new around here.

I'm trying to make an iOS app using libpd too, and wanted to make a rough and ready output limiter. Jwif's code was very useful. I made a slightly improved version (I think) that basically divides the output by the peak power - assuming it's about 3db greater than the rms. Attached.

http://www.pdpatchrepo.info/hurleur/slightly_cleaner_limiter.pd