Resotune II

[JohnRoberts]

I would love to get some drummer college student with programming chops interested, but I will repeat at this point I am tired of this project so not enthusiastic about developing a new DSP based platform. I had a prototype working using a microchip dsPic. That FFT worked well enough for a proof of concept that I could resolve the resonances quickly..(the pinout of the DSPic was close enough to the 16 bit micro I was using for me to get it running on a stock PCB). I hit a dead end coding the FFT transform to extract phase information also. I believe it can be done, just not by me easily. I managed to resolve pitch and amplitude, just not phase.

I could make a drum tuner that is faster and better, BUT it would be relatively large (two speakers), and not cheap. As long as I am doing extra tricks I might as well use a fancy graphics display (or sync to a smart phone display).

or not..

JR

 

[Tubetec]

Theres a few strobe tuner apps out there , a dual strobe display might be a good way to visually represent whats going on in Resotune ,
I found a simple one before called Tuna but it seems to have changed from what I remember ,
Certain required components and libraries might already be on the Github plate , just waiting to be forked ,

Do you happen to have any resotune II cases left over John ? I might take one if possible .
adding the guts of a modern smart speaker might be all thats needed ,
bluetooth or usb connectivity back to pc or mobile phone .

There is people here who have coded up VST/DSP type softwares ,
Whats the name of that guy , I think he's Greek , he's mainly into programming ,
sells VSTs , maybe he could help , Codemonkey or something smokes a cigarete in his picture ,

I think the Resotune idea is too good to be allowed go ,
theres many here I suspect regret not buying one while they were available ,
I know Im one ,

It needent cost a fortune , JBL do quite a nice range of mini pill type media speakers , IP splash proof rated ,
A custom mounting bracket would be easily 3D printed , a pair of drum sticks could slot in as the rail that sits on the drum hoop .

Maybe instead of worrying about the hardware end of it let that go ,
create a paid software app incorporating the principles and a custom mounting bracket for which ever speaker turns out best suited to the job ,

Its as hands free as you choose to make it , but maybe the trip down to the post office every few days is a point of contact worth keeping in your life , both with the people in your locality and customers ,

I doubt theres any need to put it to a vote , resotune deserves to go on in time ,
if anyone sends out cease and desist letters or patent infringment notices
what does it matter in the long game ?

 

[JohnRoberts]

don't under estimate the complexity involved.

I have an inventory of parts I held onto in case of warranty claims, but so far over two years and zero warranty claims, so one more year to go to fulfill my warranty obligation. After that I may give you some spare old parts while I do not want to encourage you.

I am reluctant to throw more time and money into this drummer charity.

JR

 

[Tubetec]

Using an off the shelf speaker/mic gets you away from the responsibility of having to stock spares in the future , the customer provides the electronic hardware , thats their own concern ,
Software has an EULA attached , you could easily limit your liabillity or responsibilities down the line ,

Apple I believe have a developer program , its basically aimed at students that want to try and make apps , either paid or for free , it costs a small amount yearly to join , but they basically handle the approval and distribution for you , of course they take a cut , a big cut , but its maybe the only way into the Apple phone app market place , Android is different , more like the PC in terms of what software you choose to run , chances are theres also ways to sell direct and keep most of the profits in hand , alternatively there are some very sucessful audio programs these days , Reaper and REW are basically donation based , but you get full functionality for free either way , your leaving it upto the goodwill of the users if they contribute or not , maybe its a better principle .

Again dont count on it as just doing drummers a favour , anyone interested in recording ,engineers, and producers want good sounding drums too.
As we know it can take a day or more to tune a drumkit before a recording session ,
it usually takes the vast majority of the setting up time and it leaves the rest of the band at a loose end for hours , I think theres every chance with a bit of practice to get drum tune up down to a few hours , with repeatbillity , using the Resotune.

Maybe we have forum members with the interest and the abillity to program who might get involved , if theres a willingness to co-opperate why should it really cost anything other than time to try it .

 

[john12ax7]

Are you doing all the DSP code by hand? Things like Matlab / Octave can make it much easier.

 

[user 37518]

Are you doing all the DSP code by hand? Things like Matlab / Octave can make it much easier.

They are useless when it comes to inserting the code into a DSP, neither Matlab nor Octave can create stand-alone stuff, they take up too many resources to run. Maybe something like a Raspberry Pi would be able to run Octave in it, but an 8 or 10 bit DSP, no way. Unless you mean like designing filters in Matlab/Octave just to get the coefficients and stuff like that, that is possible, but the entire routines, like the fft routine in Matlab/Octave, no way.

 

[user 37518]

Are you doing all the DSP code by hand? Things like Matlab / Octave can make it much easier.

They are useless when it comes to inserting the code into a DSP, neither Matlab nor Octave can create stand-alone programs or executables, also, Matlab and Octave take up too many resources to run. Maybe something like a Raspberry Pi would be able to run Octave in it, but an 8 or 10 bit DSP, no way. Unless you mean like designing filters in Matlab/Octave just to get the coefficients and stuff like that, that is possible, but the entire routines, like the fft routine, no way. In the end, you have to run C (or similar) in a DSP.

 

[john12ax7]

They are useless when it comes to inserting the code into a DSP, neither Matlab nor Octave can create stand-alone programs or executables

Correct me if I'm wrong, but I remember Matlab being able to interface or output c/c++ code, which while maybe not ideal, could potentially save significant development time.

 

[user 37518]

Correct me if I'm wrong, but I remember Matlab being able to interface or output c/c++ code, which while maybe not ideal, could potentially save significant development time.

You are correct, MATLAB is capable of outputting C code, but it is not very efficient since it is more of a hack rather than a one-to-one translation of m to C, and, in my experience, it rarely works flawlessly. Also, remember that many DSPs and Microcontrollers are C/C++ based, but not exactly C/C++, they have many proprietary instructions, registers, and might not be able to recognize every C instruction, and other gotchas like that. So we are talking about translating from m to C to whatever the DSP uses, that is a good way to get a buggy and inefficient code.

Long time ago I had high expectations on "making C easier" by doing stuff on Matlab and converting it to C, my hopes died very quickly since it is probably the worst way to program a C code. I don't know if this has improved over the years, but, last time I checked, it was very crappy. There are several snippets/routines for C that one can use, I have a book on my shelf called "Numerical Recipes" which is basically a numerical methods book with routines for C, you can find in it C routines for the FFT, the Sine/Cosine Transforms and other useful stuff. I believe that is one of the best alternatives.

Also, many DSPs/Microcontrollers (I mention Microcontrollers since many of them these days are so powerful that there is no need for a DSP) have the ability of programming chunks of Assembler code inside of a C code, this allows for an extremely efficient code. You can program the majority of your code in C, and only have those very critical parts relying on strict timing running on Assembler. That is practically impossible with MATLAB, and I wouldn't want to go into the mess of adapting a MATLAB to C code to do that kind of stuff. I personally know of no one using the m to C compiler, usually, they all say the same: it is a piece of crap.

That being said, DSP and Programming are not my forte, but I have accumulated a lot of mileage on MATLAB to know that it is a great tool, but terribly inefficient and resource hungry. To be honest, regardless of how many new stuff Mathworks adds to MATLAB to try to make codes transportable, the truth is that m codes only work with MATLAB installed in a computer.

 

[JohnRoberts]

I was coding using machine language. I'm self taught and it seemed more literal to understand.

The dsPic microprocessor family has several DSP friendly processor instructions like multiply/accumulate(?) to make running DSP operations simpler/faster.

I would lift the basic microcode from C dsp libraries, when I could. As I already shared I was able to read the pitch and amplitude from an audio sample using FFT, but I stalled at trying to read the phase at the same time. I found the trig identity to do it in FFT that involved using something like the arctan(?) trig function, but not how to perform that trig function in microcode. There was no stock FFT library example for reading pitch, amplitude, "and" phase shift.
===
My drum tuning invention compares the phase of the sine wave I am driving the drum head with, and the phase of the sine wave coming back. If the return is leading or lagging the driving signal that tells me that the lug is either sharp or flat and which direction to tune the lug to realize precise clear (in perfect agreement with the other lugs). I can simply measure phase without using FFT, for one sine wave at a time but since I was already running FFTs using a spread of multiple sine waves simultaneously why not grab the phase of each sine at the same time?

But like i said I pretty much lost interest in designing an improved RESOTUNE drum tuner several years ago.

I haven't checked the dsp libraries for several years now.

JR

PS; In hindsight I do not have to use FFT to measure phase to clear drum lugs. That was an artificial road block I set up for myself. I still don't feel like going back down that coding rabbit hole, now that I no longer have that excuse.

 

[Tubetec]

John Mulcahey inventor of REW likely knows a thing or two on the subject ,