How to detect gesture as fast as possible?

Thread Starter

goutham1995

Joined Feb 18, 2018
94
Is there any way to identify which finger I have used while hitting a drum? I had thought of using fingerprint sensor but it's slow for my application. I need it to be as fast as recognising in 20-40 ms. And it needs to be sensitive too
 

ebeowulf17

Joined Aug 12, 2014
3,275
Is there any way to identify which finger I have used while hitting a drum? I had thought of using fingerprint sensor but it's slow for my application. I need it to be as fast as recognising in 20-40 ms. And it needs to be sensitive too
Are you willing to paint your fingers different colors for easy recognition? If not, nothing else comes to mind!

What are you trying to accomplish with the recognition of different fingers? Presumably there's some higher goal, and finger identification is an intermediate step. Maybe there's a more effective solution to your higher goal that won't require finger ID.
 

BR-549

Joined Sep 22, 2013
4,938
Video it. Then single step frame it. Might need higher speed frame rate.

Try taping a weight on finger in question for real time acknowledgement.
 

DickCappels

Joined Aug 21, 2008
6,957
This is not the kind of question ordinary people ponder in their idle moments, so don't be disappointed that we don't have ready answers. My first response is why not have areas on the drum-head dedicated to certain fingers, much as keys on a keyboard are dedicated to particular fingers?

Additionally, the time it takes to recognize a fingerprint is highly dependent upon both the algorithm and the processing power involved. If you can simplify the algorithm you can speed up the process for a given processing environment. Consider that you are only in need of discriminating one finger from another among ten possible fingers of an individual person rather than the 76 billion human fingers on earth.
 
Lots and lots of wired gloves for many purpose. Perhaps you can find one that is both economical and applicable to your needs.

Search here.

And check this blog where Jen is getting into some of the nitty gritty of building one.
 

MrChips

Joined Oct 2, 2009
22,878
Raymond beats me to it.
I was going to say wear a glove that has elastomeric resistive or capacitive sensors at the finger tips.
 

Thread Starter

goutham1995

Joined Feb 18, 2018
94
Raymond beats me to it.
I was going to say wear a glove that has elastomeric resistive or capacitive sensors at the finger tips.
I had thought of gloves earlier but my professor was against it. We're trying to make an electronic version of Indian classical musical instrument called mridangam and we want the player to have a natural feel about it while playing.
 

Thread Starter

goutham1995

Joined Feb 18, 2018
94
Are you willing to paint your fingers different colors for easy recognition? If not, nothing else comes to mind!

What are you trying to accomplish with the recognition of different fingers? Presumably there's some higher goal, and finger identification is an intermediate step. Maybe there's a more effective solution to your higher goal that won't require finger ID.
I'm trying to make an electronic version of an Indian percussion instrument called mridangam. The problem is, there are two notes which are played on the same spot, one with little finger and the other one with index finger. The sounds produced are different in the actual instrument. I had tried taking a pvc disc and hit it the same way I hit on actual instrument hoping the frequencies of vibration will be different.. But the accuracy was only 70-80 percent. It has to be close to 100 percent. So, if I can find out which finger I'm using, that can help
 
I had thought of gloves earlier but my professor was against it. We're trying to make an electronic version of Indian classical musical instrument called mridangam and we want the player to have a natural feel about it while playing.
Well, that might increase the task difficulty a bit. Color sensors have already been mentioned. What about accelerometers mounted above the metacarpals. The finger tips and much of the finger could remain pristine. Of course that is not a simple program to write.
 

BR-549

Joined Sep 22, 2013
4,938
He's trying to synthesize an instrument with a real instrument? You mean he wants to electrify it?

Ok.....will the hands always be at the same orientation to circumference?..............or do you need to rotate without change in sound?

Do you understand my question? We need to know possible sensor area.
 

ebeowulf17

Joined Aug 12, 2014
3,275
I'm trying to make an electronic version of an Indian percussion instrument called mridangam. The problem is, there are two notes which are played on the same spot, one with little finger and the other one with index finger. The sounds produced are different in the actual instrument. I had tried taking a pvc disc and hit it the same way I hit on actual instrument hoping the frequencies of vibration will be different.. But the accuracy was only 70-80 percent. It has to be close to 100 percent. So, if I can find out which finger I'm using, that can help
If I were you, I'd focus on what's different about those fingers, and how that translates into tone/timbre changes. Is one finger skinny and hard, and the other big and meaty? Does one naturally have a tendency to bounce away from the head more quickly, allowing it to resonate longer, while the other inadvertently lingers a bit, muting the head more quickly? Is one naturally hitting with the pad/fingerprint area while the other hits more with the end of the finger? There must be a reason it sounds different, and l wouldn't be at all surprised if it was based on technique as much as it is on any physical characteristic of the fingers.

I've done a lot of work as an audio engineer, a lot of playing drum kits, but only a small amount of hand drumming, so I can't tell you exactly what to look for. However, I'm fairly optimistic that if you really think about WHY that finger makes a different musical sound, you'll come up with a better electronic means of recognizing it.

What do you plan to have the player striking? Is a PVC disc the ultimate goal? Have you considered silicone pads like many practice pads and electronic drum kits use? What did you have in mind for pickups? Piezo sensors, pressure transducers, capacitive, optical, the possibilities are endless...

There are existing electronic drums (both kits for sticks and hand drums) so I'm sure many of these hurdles have been figured out. If any of that information is publicly available, it might give you something to build on.
 

MrChips

Joined Oct 2, 2009
22,878
If I understand this correctly, TS is trying to create an electronic mridangam, using techniques similar to an electronic drum kit, only in this case the use of specific fingers is essential.
 

BR-549

Joined Sep 22, 2013
4,938
The only feasible thing I can think of.....and I could not do......is modify a speech recognition program. Might not even have to modify it......I'm sure they are better at it now.

Just teach it your "hand"......and go to town.

Are you allowed to use software?
 

Thread Starter

goutham1995

Joined Feb 18, 2018
94
The only feasible thing I can think of.....and I could not do......is modify a speech recognition program. Might not even have to modify it......I'm sure they are better at it now.

Just teach it your "hand"......and go to town.

Are you allowed to use software?
you mean use ML?
 
Top